The significance of the opioid epidemic has led to increasing controlled drug oversight from regulatory agencies with the expectation that controlled drugs are closely tracked to monitor for diversion. The Code of Federal Regulations Title 21 Chapter II has extensive controlled substance requirements, including many that require Drug Enforcement Administration (DEA) registrants to closely monitor their controlled drugs for diversion and report any significant discrepancies. The overarching theme appears in part 1301.71, which states, “All applicants and registrants shall provide effective controls and procedures to guard against theft and diversion of controlled substances.”1 The regulations go on to elaborate on this concept, including part 1304.21, which requires keeping a complete and accurate record of all controlled substances,2 and part 1301.76, which requires notifying the DEA of any significant theft/loss.3
In the case US versus Little, it is clear that proven controlled substance discrepancies (shortages or overages) without appropriate paperwork reporting or rectifying them constitute recordkeeping violations.4 These recordkeeping violations can incur fines of up to $10,000 per violation.5 The seriousness of these violations and fines supports the need for DEA registrants to have a reliable process for monitoring controlled substances on hand and addressing any discrepancies. It is relatively straight forward to monitor the amount remaining in full vials/bottles and to count tablets/capsules remaining in open bottles. However, the ideal method for monitoring quantities remaining in partial multidose vials is not clear.
In human hospitals, the Joint Commission for Healthcare, which is the accreditation body, provides significant restrictions regarding the continual use of multidose vials due to contamination concerns.6 The result, as it pertains to controlled substances, is much less need to continually monitor the amount remaining in open vials of controlled substances in the human hospital setting. In contrast, veterinary medicine uses multidose, and often single-use, vials of controlled substances for extended periods or until gone.7
In an attempt to provide sufficient oversight of controlled drugs on hand and identify quantity discrepancies in a timely manner while continuing to use vials in a multidose fashion until gone, the veterinary profession has adopted various methods, each with pros and cons. When considering the quality of these methods, accuracy and precision are both important. In general, accuracy is how closely the results obtained align with the true result, and precision is how closely the results obtained align with each other. An ideal method will be both accurate and precise.
The primary objective of this study was to evaluate the precision and accuracy of 3 common methods for monitoring volumes in vials of injectable controlled drugs. The methods evaluated were:
Actual draws of the volume remaining.
Weight tracking of the volume remaining and/or the volume removed.
Discrepancy percentage at the end of each vial.
The secondary objective was to identify factors that affected the accuracy and precision of the methods evaluated. We hypothesized that the accuracy and precision of methods 1 and 2 would be similar to each other and more precise than method 3.
Methods
Actual draws of the volume remaining (method 1)
To mimic an injectable drug vial with overfill of less than 10%, a 30-mL empty sterile vial was filled with 10.5 mL of sterile water for injection. A series of doses was drawn from the vial using either a 1-mL syringe with 0.01-mL markings (for dose volumes less than 0.75 mL) or a 3-mL syringe with 0.1-mL markings (for dose volumes greater than 0.75 mL) and a 25-gauge, 1-inch needle. Volumes removed were based on standard volumes of methadone used for small animal patients at the teaching institution where this study was completed. For a full list of dose volumes drawn up, see Table 1. After each dose was removed, a larger syringe appropriate for the volume expected to remain was used with a 22-gauge, 1-inch needle to withdraw the remaining quantity of liquid. The remaining volume was noted, and the liquid was returned to the vial.
Volumes in milliliters removed to simulate doses for actual draws of the volume remaining (method 1) and weight tracking of the volume remaining and/or the volume removed (method 2).
| Volume removed for each simulated dose for methods 1 and 2 (in mL) | Dose volume summary statistics (in mL except count) | |
|---|---|---|
| 0.01 | Mean | 0.20 |
| 0.04 | Median | 0.14 |
| 0.04 | Minimum | 0.01 |
| 0.05 | Maximum | 0.90 |
| 0.11 | Count | 33 |
| 0.11 | Mean | 0.20 |
| 0.11 | ||
| 0.12 | ||
| 0.12 | ||
| 0.12 | ||
| 0.12 | ||
| 0.12 | ||
| 0.12 | ||
| 0.14 | ||
| 0.14 | ||
| 0.14 | ||
| 0.14 | ||
| 0.15 | ||
| 0.15 | ||
| 0.17 | ||
| 0.17 | ||
| 0.18 | ||
| 0.18 | ||
| 0.18 | ||
| 0.18 | ||
| 0.18 | ||
| 0.18 | ||
| 0.25 | ||
| 0.4 | ||
| 0.4 | ||
| 0.54 | ||
| 0.76 | ||
| 0.9 | ||
In this study, method 1 was compared to method 2 and discrepancy percentage at the end of each vial (method 3) to compare the accuracy and precision of each method for monitoring volumes in vials of injectable controlled drugs. Evaluating methods 1 and 2 required removing predetermined volumes from the vial and then using the method being evaluated to determine the amount remaining. These predetermined volumes were designed to mimic volumes of controlled substances likely used in a small-animal veterinary practice. For both methods, the volumes were removed from smallest to largest as shown in this table. Summary statistics for the volumes removed are also provided.
Expected hub loss was calculated as 0.05 mL/stick based on common practice at the teaching institution where this study was completed. Actual hub loss that occurred was calculated based on volume remaining versus volume expected. Calculations were completed based on the 10.5-mL known start volume as well as a 10-mL start volume to mimic common practice more closely where exact overfill amount is unknown. Precision for this method was defined as the amount of deviation of hub loss in milliliters per stick for each actual draw compared to the mean hub loss in milliliters per stick. Accuracy for this method was defined as the amount of deviation of actual hub loss in milliliters per stick for each actual draw compared to the expected hub loss of 0.05 mL/stick. Absolute value was used for all calculations.
Weight tracking of the volume remaining and/or the volume removed (method 2)
A 30-mL empty sterile vial was filled with 10 mL of sterile water for injection. A series of doses were drawn from the vial using either a 1-mL syringe with 0.01-mL markings (for dose volumes less than 0.75 mL) or a 3-mL syringe with 0.1-mL markings (for dose volumes greater than 0.75 mL) and a 25-gauge, 1-inch needle. Volumes removed were based on standard volumes of methadone used for small animal patients at the teaching institution where this study was completed. For a full list of dose volumes drawn up, see Table 1. The syringe with the needle attached was weighed immediately prior to each draw and reweighed after the draw. The vial was weighed after each draw. The weights were recorded on 2 scales with differing readability. Scale 1 (Mettler Toledo model XS105DU; Mettler-Toledo) had a max weight of 120 g and a readability of 0.01 mg. Scale 2 (Adventurer Pro AV812C; OHAUS) had a max weight of 810 g and a readability of 0.01 g.
The actual weight of the volume in grams per milliliter was calculated and compared to the expected weight of 1 g/mL. The expected weight was based on the density of water. Precision for this method was defined as the amount of deviation of grams per milliliter weight for each measurement compared to the mean grams per milliliter weight. Accuracy for this method was defined as the amount of deviation of grams per milliliter weight for each measurement compared to 1 g/mL (density of water). The differences between the accuracy and precision based on scale and syringe versus vial weights were also calculated. Absolute value was used for all calculations.
Discrepancy percentage at the end of each vial (method 3)
The total discrepancy volume per vial was determined for each methadone vial in the intermediate care and ICU wards of the teaching institution where this study was completed from January 2022 through March 2023. The vials used in these locations were repackaged in a cleanroom facility by equally splitting the total volume in a 20-mL commercially available vial of methadone into 2 approximately 10-mL vials. Because the commercial vial contained a small amount (< 10%) of overfill, the repackaged vials were assumed to each contain a small amount of overfill, but the exact overfill amount was unknown. The percentage of discrepancy between the calculated amount remaining (10 mL − total dose volume) and the 0 mL actually remaining was determined for each vial with the absolute value used for the precision calculation. Precision for this method was defined as the amount of deviation of discrepancy percentage for each measurement compared to the mean discrepancy percentage. An accuracy calculation was not applicable for this method.
Statistical analysis
Excel (Microsoft Corp) was used for all data analysis. Summary data was calculated for the accuracy and precision for each method based on the definitions explained above. Linear regression was used to evaluate the effect of dose volume for methods 1 and 2. For method 2, linear regression was also used to evaluate the effect of scale readability, and a 2-sided t test was performed to evaluate the effect of weighing the syringe versus the vial on accuracy and precision. For method 3, linear regression was used to evaluate the effect of number of sticks on the precision. Significance for all tests was set at P < .05. A 2-tailed t test was used to compare the mean accuracy and precision across each method.
Results
For method 1, actual draw accuracy based on a 10-mL base and a 10.5-mL base was 21.79% and 67.26%, respectively. Actual draw precision based on a 10-mL base and a 10.5-mL base was 95.81% and 99.03%, respectively. Summary data for method 1 accuracy and precision is shown in Table 2. Accuracy and precision data for the 10.5-mL base were used for comparison with the other methods. Actual hub loss per stick based on a 10-mL base and a 10.5-mL base was 0.01 mL and 0.05 mL, respectively. Summary data for hub loss is shown in Table 3, with graphical representations in Figure 1.
Accuracy and precision summary data by method and variation.
| Accuracy | |||||||
| Method | Actual draws (method 1) | Weights (method 2) | |||||
| Variation 1 | 10-mL base | 10.5-mL base | Vial | Syringe | |||
| Variation 2 | N/A | Scale 1 | Scale 2 | Scale 1 | Scale 2 | ||
| Definition | The amount of deviation of actual hub loss in milliliters per stick for each actual draw compared to an expected hub loss of 0.05 mL/stick. | The amount of deviation of grams per milliliter weight for each measurement compared to 1 g/mL (density of water). | |||||
| Mean (%) | 21.79 | 67.26 | 98.14 | 98.14 | 98.53 | 98.52 | |
| Median (%) | 13.00 | 63.00 | 98.48 | 98.50 | 98.68 | 98.67 | |
| Minimum (%) | 0.00 | 49.00 | 90.79 | 91.00 | 94.38 | 95.00 | |
| Maximum (%) | 64.00 | 100.00 | 99.05 | 99.04 | 99.06 | 99.06 | |
| Spread (%) | 64.00 | 51.00 | 8.26 | 8.04 | 4.68 | 4.06 | |
| Precision | |||||||
| Method | Actual draws (method 1) | Weights (method 2) | Discrepancy percentage (method 3) | ||||
| Variation 1 | 10-mL base | 10.5-mL base | Vial | Syringe | N/A | ||
| Variation 2 | N/A | Scale 1 | Scale 2 | Scale 1 | Scale 2 | N/A | |
| Definition | The amount of deviation of hub loss in milliliters per stick for each actual draw compared to the mean hub loss in milliliters per stick. | The amount of deviation of grams per milliliter weight for each measurement compared to the mean grams per milliliter weight. | The amount of deviation of discrepancy percentage for each measurement compared to the mean discrepancy percentage. | ||||
| Mean (%) | 95.81 | 99.03 | 99.00 | 99.00 | 99.00 | 99.00 | 60.25 |
| Median (%) | 95.99 | 99.10 | 99.18 | 99.19 | 99.11 | 99.10 | 71.53 |
| Minimum (%) | 88.49 | 97.92 | 95.05 | 95.15 | 96.19 | 96.62 | 0.00 |
| Maximum (%) | 99.27 | 99.20 | 99.49 | 99.48 | 99.36 | 99.36 | 98.09 |
| Spread (%) | 10.78 | 1.28 | 4.44 | 4.33 | 3.17 | 2.74 | 98.09 |
N/A = Not applicable.
As described in Table 1, 3 methods were evaluated for accuracy and precision when tracking controlled drug volumes remaining in multidose vials. To analyze for confounding variables, variations were evaluated for methods 1 and 2. For method 1, the vial was known to contain 10.5 mL of sterile water to mimic the presence of overfill that would likely occur with a vial labeled to contain 10 mL of drug. For this method, the accuracy and precision were both evaluated based on a starting volume of 10.5 mL (actual known volume) and 10 mL (expected volume when mimicking clinical practice). Accuracy and precision summary statistics for method 1 are shown for both starting volumes. For method 2, both the vial (volume remaining) and the syringe (volume removed) were weighed, and 2 different scales with different maximum weighable quantities and readabilities were used, with scale 1 representing the scale with a lower maximum weighable quantities and higher readability. Accuracy and precision summary statistics are shown for both vial and syringe weights for both scales. For method 3, accuracy did not apply and there were no variations, so only 1 set of summary statistics is shown for this method.
Actual hub loss per stick summary data for methods 1 and 3.
| Actual hub loss per stick | |||
|---|---|---|---|
| Actual draws (method 1): 10-mL base | Actual draws (method 1): 10.5-mL base | Discrepancy percentage (method 3) | |
| Mean (mL) | 0.01 | 0.05 | 0.04 |
| Median (mL) | 0.02 | 0.04 | 0.04 |
| Minimum (mL) | −0.07 | 0.04 | 0.00 |
| Maximum (mL) | 0.04 | 0.10 | 0.13 |
| Spread (mL) | 0.11 | 0.06 | 0.13 |
As described in Table 1, 3 methods were evaluated for accuracy and precision when tracking controlled drug volumes remaining in multidose vials. As described in Table 2, variations of these methods were included to analyze for confounding factors. Hub loss was known to occur with the methods requiring drawing volumes (methods 1 and 3), and an estimate of 0.05 mL was used for the purposes of the study. However, the actual hub loss experienced with the use of 22-gauge, 1-inch needles and 1-mL (for volumes less than or equal to 0.75 mL) or 3-mL (for volumes greater than 0.75 mL) syringes was calculated. The summary statistics are shown in this table.
Actual hub loss in milliliters per vial puncture that occurred during actual draws of the volume remaining (method 1). In this study, method 1 was compared to weight tracking of the volume remaining and/or the volume removed (method 2) and discrepancy percentage at the end of each vial (method 3) to compare the accuracy and precision of each method for monitoring volumes in vials of injectable controlled drugs. For method 1, the vial was known to contain 10.5 mL of sterile water to mimic the presence of overfill that would likely occur with a vial labeled to contain 10 mL of drug. Therefore, actual hub loss was calculated based on both the 10.5 mL (represented by circles) and 10 mL (represented by diamonds) volumes. As shown in this figure, regardless of the assumed starting volume, the actual hub loss in milliliters per stick converged on 0.04 mL. This was based on the use of 22-gauge, 1-inch needles and 1-mL (for volumes less than or equal to 0.75 mL) or 3-mL (for volumes greater than 0.75 mL) syringes.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.24.10.0307
For method 2, as shown in Table 4, no statistical difference was found for vial weights versus syringe weights. There was also no statistical difference found for weights on scale 1 compared to scale 2. The mean accuracy and percentage were 98.14% and 99%, respectively. Full summary data for method 2 accuracy and precision is shown in Table 2.
Effect of dose volume, total volume removed, number of sticks, and syringe-versus-vial weights on accuracy and precision.
| Effect of dose volume on accuracy and precision | Effect of number of sticks | ||||
|---|---|---|---|---|---|
| Weight tracking (method 2) syringe | Weight tracking (method 2) vial | Discrepancy percentage (method 3) | |||
| Accuracy | Precision | Accuracy | Precision | Precision | |
| R2 | 0.14 | 0.14 | 0.14 | 0.14 | 0.004 |
| P value | .04a | .03a | .03a | .03a | .59 |
| Effect of total volume removed on accuracy and precision | |||||
| Actual draws (method 1): 10-mL base | Actual draws (method 1): 10.5-mL base | ||||
| Accuracy | Precision | Accuracy | Precision | ||
| R2 | 0.57 | 0.22 | 0.59 | 0.13 | |
| P value | .000000344a | .006a | .0000002a | .04a | |
| Effect of syringe vs vial on scale accuracy | |||||
| Effect of syringe vs vial weights | Effect of scale 1 vs scale 2 (vial weights) | Effect of scale 1 vs scale 2 (syringe weights) | |||
| P value | .18 | .99 | .97 | ||
Value is statistically significant.
As described in Table 1, 3 methods were evaluated for accuracy and precision when tracking controlled drug volumes remaining in multidose vials. As described in Table 2, variations of these methods were included to analyze for confounding factors. Linear regression was used to evaluate the effect of dose volume for methods 1 and 2. For method 2, linear regression was also used to evaluate the effect of scale readability, and a 2-sided t test was performed to evaluate the effect of weighing the syringe versus the vial on accuracy and precision. For method 3, linear regression was used to evaluate the effect of number of sticks on the precision. Significance for all tests was set at P < .05. The R2 value and associated P value are shown for each.
For method 3, data from 82 vials was obtained, and the discrepancy percentage of mean precision was 60.25%. Summary data for method 3 precision is shown in Table 2. Accuracy was not applicable for this method.
The effect of volume removed on accuracy and precision was evaluated for methods 1 and 2. Dose volume was positively correlated with both accuracy and precision for weights (method 2). Total volume removed was positively correlated with both accuracy and precision for actual draws (method 1). The effect of number of sticks per vial was not correlated with discrepancy percentage (method 3) of precision. R2 and P values are shown in Table 4.
The mean accuracy for weighing drug was significantly higher than mean accuracy for actual draws (10.5-mL base). Mean precision was not statistically different between these 2 methods. Mean precision was significantly higher for both weight tracking and actual draws (10.5-mL base) than for discrepancy percentage. Mean accuracy and precision comparison P values are shown in Table 5.
Mean accuracy and precision comparison across methods.
| Comparison of accuracy means across methods (2-tailed t test P values) | Comparison of precision means across methods (2-tailed t test P values) | |||
|---|---|---|---|---|
| Actual draws (10.5-mL base) | Actual draws (10.5-mL base) | Discrepancy percentage | ||
| Weight tracking | 6.952−126a | Actual draws (10.5-mL base) | N/A | 1.27−10a |
| Weight tracking | 0.73 | 1.32−10a | ||
Value is statistically significant.
As described in Table 1, 3 methods were evaluated for accuracy and precision when tracking controlled drug volumes remaining in multidose vials. As described in Table 2, variations of these methods were included to analyze for confounding factors. A 2-tailed t test was used to compare mean accuracy and precision across each method, with only 1 variation selected for each method. Significance for all tests was set at P < .05. The P values are shown for each.
Discussion
The results of this study indicate that method accuracy in decreasing order is weight tracking of the volume remaining and/or the volume removed (method 2), actual draws of the volume remaining (method 1), and discrepancy percentage at the end of each vial (method 3). Method precision is equal for methods 1 and 2, with method 3 having the lowest precision.
This study demonstrated a significant impact of overfill on the accuracy of method 1, with this method being significantly more accurate when the accuracy was based on a starting volume of 10.5 mL, which took the overfill into account. However, taking overfill into account only increased precision slightly for this method, which is to be expected based on how accuracy and precision were defined. A potential concern with this finding is that overfill is not typically known in practice, so diversion of the overfill may occur without being noticed.
Several factors were evaluated to determine if they contributed to the accuracy and/or precision of each method. For method 1, the total volume previously removed was shown to be positively correlated with the accuracy and precision of this method. This was an expected association as the smaller the amount remaining at the time of the actual draw, the smaller the syringe that could be used, resulting in more accurate and precise measurements. Practices utilizing actual draws to monitor drug remaining should keep in mind that this measurement is more likely to be of benefit when the vial has less drug remaining.
For method 2, the volume of the dose removed was positively correlated with the accuracy and precision of this method. However, no difference was seen between vial and syringe weights nor between scale 1 versus scale 2. For the finding that method 2 has improved accuracy and precision with larger volumes, this has clinical implications because this method may be a less reliable option for drugs such as methadone, where very small volumes are frequently used. The lack of significant difference between the 2 scales studied was unexpected due to the difference in scale resolution (0.01 mg vs 0.01 g). However, this finding supports that more economical scales may be an option for practices wishing to utilize this method. The results of this study also support that either a process weighing the syringe with the dose removed or a process weighing the quantity remaining in the vial can be used with comparable accuracy and precision.
For method 3, the number of vial punctures per vial did not affect the precision of this method. Therefore, this method can be expected to perform similarly regardless of how many sticks are required to utilize the full vial contents.
When utilizing actual hub loss determined through method 1 (with and without considering overfill), the actual hub loss for both converged at 0.04 mL/stick. This is based on the use of a 22-gauge, 1-inch needle and the smallest syringe size to accurately remove the volume. It is expected that if a larger gauge or larger needle is used, this hub loss may increase slightly, and consistent use of a hub-less syringe may decrease the actual hub loss. This result supports the use of 0.05-mL/stick hub loss estimate in practice settings that routinely use similar needle and syringe sizes utilized in this study.
This study did have several limitations. First, accuracy definitions were based on the assumption that hub loss would be 0.05 mL/stick, which was an estimated amount used at the study institution, and it is known that actual hub loss may vary with changes in average needle and syringe size as well as with the drug used (eg, euthanasia solution will likely have additional hub loss due to higher density compared to methadone). Since the DEA does not provide a standard hub loss calculation amount, each institution should determine an appropriate estimate based on syringe size, needle size, and specific drugs used in that practice. Second, this study was conducted with sterile water to mimic the majority of injectable controlled drugs that have densities close to that of water. However, these results may not accurately extrapolate to thicker drugs, such as euthanasia solution. Third, the study draws and weights were conducted by a single, trained individual, so results in practice may vary based on the number of individuals involved and the training of those individuals.
In addition to the precision and accuracy, each method has pros and cons to use. Method 1 utilizes a process that most people are familiar with, does not require a readily available scale, does not require density considerations, and allows thinking in volume, which makes math easier. However, repeated vial sticks have the potential to introduce contamination and decrease vial stopper integrity, leading to leaking and resulting in additional hub loss. These limitations for method 1 can be significantly decreased by the use of a closed system vial adaptor that allows only spiking the vial once and then conducting all subsequent draws with a Leur-lock connection. Method 2 prevents the disadvantages of method 1 but requires a readily available scale at drug storage locations, requires that correct scale use be taught, math is required to think in volume, the density of the drug must be known, or the starting weight of a full vial and the empty weight of the vial must be known for each drug, manufacturer, and vial size use. Method 3 does not require any tracking while the vial is in use, making it easier and less time consuming, and it only requires basic math that can be done when time allows. However, with this method, diversion can occur without being noticed; it relies on trends, so initial problems are less likely to be detected; and it is difficult to determine when a problem occurred during the use of a vial. The method(s) chosen by a specific practice should be based on accuracy and precision as well as the pros and cons of each method. In conclusion, weight tracking and actual draws are more robust methods than postuse discrepancy percentage review.
Acknowledgments
None reported.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the generation of this manuscript.
Funding
The authors have nothing to disclose.
ORCID
L. R. Forsythe https://orcid.org/0000-0002-4304-8464
References
- 1.↑
Code of Federal Regulations 1301.71: security requirements generally. National Archives and Records Administration. Accessed January 21, 2024. https://www.ecfr.gov/current/title-21/chapter-II/part-1301/subject-group-ECFRa7ff8142033a7a2/section-1301.71
- 2.↑
Code of Federal Regulations 1304.21: general requirements for continuing records. National Archives and Records Administration. Accessed January 21, 2024. https://www.ecfr.gov/current/title-21/chapter-II/part-1304/subject-group-ECFRed220d26113f6a5/section-1304.21
- 3.↑
Code of Federal Regulations 1301.76: other security controls for practitioners. National Archives and Records Administration. Accessed January 21, 2024. https://www.ecfr.gov/current/title-21/chapter-II/part-1301/subject-group-ECFRa7ff8142033a7a2/section-1301.76
- 4.↑
United States v. Little, 59 F. supp. 2d 177 (D. Mass. 1999). Justia. Accessed January 23, 2024. https://law.justia.com/cases/federal/district-courts/FSupp2/59/177/2408545/
- 5.↑
US Code 842: prohibited acts B. Govregs. Accessed January 21, 2024. https://www.govregs.com/uscode/expand/title21_chapter13_subchapterI_partD_section842#uscode_1
- 6.↑
Multi-dose vials - managing multi-dose vials of injectable medication. The Joint Commission. Accessed January 21, 2024. https://www.jointcommission.org/standards/standard-faqs/home-care/medication-management-mm/000001529/
- 7.↑
Sabino CV, Weese JS. Contamination of multiple-dose vials in a veterinary hospital. Can Vet J. 2006;47(8):779–782.
