Introduction

Opportunities have been identified regarding timely delivery of medications due to lack of ADC inventory related to stock outs or other variances in pharmacy workflow (i.e. compounding and distribution times). These opportunities for improvement impact patient care and result in both nursing and pharmacy staff frustration. Additionally, there are significant costs associated with unused medications stored in ADCs, taking up valuable real estate that would otherwise house more opportune inventory. This situation has created a need for more efficient management of the ADC inventory. Currently, research in the topic area is sparse. We hypothesize that the systematic management and oversight of ADC inventory will demonstrate a significant improvement in key performance indicators and provide insight to the current gaps in knowledge.

Methods

This study will be a continuous quality improvement project with a combination of retrospective data review and prospective optimization interventions of automated dispensing cabinets (Pyxis®). Interventions include review and adjustment of ADC par levels, removal of unused/stagnate medications, standardization of stock, and continual review of ADC inventory turns and associated optimization opportunities. The primary outcomes are the change in vend/fill ratio from baseline, change in medication stockout percentage from baseline.

Results

There was no significant difference in vend/fill ratio after the optimization phase compared with baseline [Difference 0.13 (11.56 ± 6.1 vs. 11.43 ± 5.41) respectively, (p=0.84)]. Medication stockout percentage was also found to be similar with baseline [Difference −0.05 (0.71% ± 0.12 vs. 0.76% ± 0.08) respectively, (p=0.37)]. For secondary outcomes, the change in blind stockout percentage from baseline was −0.04 [0.13 ± 0.02 vs. 0.17 ± 0.02, (p=0.004)] and the change in medications dispensed per day from baseline was 317 [2656 ± 143 vs. 2339 ± 200, (p=0.0002)].

Conclusion

Optimization of automated dispensing cabinets yielded marginal improvements in vend/fill ratio and stockout percentage and significantly improved overall efficiency through an increase in the number of medications stocked in ADCs and number of medications dispensed per day from ADCs. Evaluation of more clinically significant performance indicators may better characterize the benefits from the optimization process.

Automated dispensing cabinets (ADC) are a common means for medication distribution in modern day hospitals.(1) ADCs have improved overall efficiency in the medication use process by storing medications directly on the patient care unit, allowing for timely medication procurement and administration. While the benefits of ADCs are widely recognized, they are not without their shortcomings. Opportunities have been identified regarding timely delivery of medications due to lack of ADC inventory related to stock outs or other variances in pharmacy workflow (i.e. compounding and distribution times) at Kaiser Permanente Vacaville Medical Center. These opportunities for improvement impact patient care and result in increases of both nursing and pharmacy staff frustration. Additionally, there are significant costs associated with unused medications stored in ADCs, taking up valuable real estate that would otherwise house more opportune inventory. This situation has created a need for more efficient management of the ADC inventory.

Currently, research in the topic area is sparse. However, two studies provide insight into the potential benefits from optimization of ADCs. A retrospective study by Lupi et al. 2019, looked at optimization of 65 ADCs at a large tertiary medical center.(2) Through management of ADC par levels, the authors saw a reduction in the amount of medications dispensed manually from the central pharmacy, a decrease in the frequency of medication stockouts on the patient care units, and were able to increase overall cabinet inventory while decreasing cost. Major limitations include exclusion of ADCs in the emergency department, post-anesthesia care units, and procedural areas; to note, ADC optimization occurred within in a short time frame of an 8-week period. The second study by O’Neil et al. 2016, was a small prospective study that yielded modest improvements in vend/fill ratio but performed optimization on only eight nonprofiled ADCs over a 6-month period.(3)

This study aims to determine the feasibility and potential benefits associated with improving ADC management through improved inventory oversight and generalizability by providing a longer period of optimization of ADCs in more patient care areas. We hypothesize that the systematic management and oversight of ADC inventory will demonstrate a significant improvement in key performance indicators and provide insight to the current gaps in knowledge.

Study Design

This study will be a continuous quality improvement project with a combination of retrospective data review and will include prospective optimization intervention on automated dispensing cabinets (Pyxis®) at Kaiser Permanente Vacaville Medical Center from March 31, 2019 to January 31, 2020.

Kaiser Permanente Vacaville is a 152-bed level 2 trauma center located in Vacaville, California. The medical center has approximately 47 automated dispensing cabinets (Pyxis®). For the purposes of this study, ADC optimization was defined as the manipulation of inventory stock and stock levels within the ADC to improve operational efficiency related to the medication distribution and use process. Optimization interventions involved: 1) monthly review and adjustment of ADC par levels to decrease refills and inventory stockouts; 2) replacement of unused/stagnant medications with more opportune medications; 3) standardization of stock defined as mirroring select medications in ADCs on the same floor; and 4) continual review of ADC inventory turns and associated optimization opportunities. These procedures to be conducted are part of standard pharmacy workflow. Study objectives will be measured and assessed on a monthly basis. The study was determined to be exempt from institutional board review. See terminology in Table 1 for more information.

Study Population

Study population will include the automated dispensing cabinets (Pyxis®) at Kaiser Permanente Vacaville Medical Center. ADCs that will be excluded are cabinets within the central pharmacy or not linked to a patient profile.

Study Endpoints

The primary outcomes are the change in vend/fill ratio from baseline and change in medication stockout percentage from baseline. Baseline period was defined as the period 3 months prior to optimization (January 1, 2019 to March 30, 2019). We will also review outcomes to include a sample of medication costs as defined by WAC pricing from RED BOOK; change in medication blind stockouts percentage from baseline; change in number of medications dispensed per day from baseline; and the number of expired medications removed during the study period.

Sources of data and data to be collected include: Kaiser Permanente Health Connect (for medications not dispensed by ADC and filled in central pharmacy), BD Knowledge Portal® (for stockout percentage, blind stockout percentage, number of medications dispensed per day, number of expired medications removed, vend/fill ratio), Permanente Online Interactive Network of Tools (POINT®) (for vend/fill ratio, stockout percentage and missing medication data for med surge floors and ICU), and BD Enterprise Server (ES) Portal® (for PAR levels, current inventory and location stocked). All information is to be extracted electronically from the above data sources without using a data collection sheet. Data is then converted into a Microsoft Excel® file to be formatted and filtered to display the information of interest as stated above.

Data will subsequently be analyzed via manual review (format of data described above). Primary and secondary outcomes will be compared using paired t-test for continuous data. Two-sided statistical power test was calculated using alpha=0.05. A p-value less than 0.05 was determined to be statistically significant.

A total of 47 automated dispensing cabinets (ADCs) met inclusion criteria and 21 ADCs were excluded leaving 26 ADCs for the study population. There was no significant difference found in the vend/fill ratio after the optimization phase compared with baseline [Difference 0.13 (11.56 ± 6.1 vs. 11.43 ± 5.41) respectively, (p=0.84)] (figure 1). Medication stockout percentage was also found to be similar with baseline [Difference −0.05 (0.71% ± 0.12 vs. 0.76% ± 0.08) respectively, (p=0.37)] (figure 2).

Figure 1.

Change in vend/fill ratio after optimization from baseline. Vend/fill ratio calculated for each automated dispensing cabinet using the equation listed in Table 1. The bars indicate standard deviation.

Figure 1.

Change in vend/fill ratio after optimization from baseline. Vend/fill ratio calculated for each automated dispensing cabinet using the equation listed in Table 1. The bars indicate standard deviation.

Close modal
Figure 2.

Change in stockout percentage after optimization from baseline. Stockout percentage calculated for each automated dispensing cabinet using the equation listed in Table 1. The bars indicate standard deviation.

Figure 2.

Change in stockout percentage after optimization from baseline. Stockout percentage calculated for each automated dispensing cabinet using the equation listed in Table 1. The bars indicate standard deviation.

Close modal

For secondary outcomes, the change in blind stockout percentage from baseline was −0.04 [0.13 ± 0.02 vs. 0.17 ± 0.02, (p=0.004)] (figure 3) and the change in number of medications dispensed per day from baseline was 317 [2656 ± 143 vs. 2339 ± 200, (p=0.0002)] (figure 4). The number of expired medications removed (through the outdating process) during the optimization period totaled 3590 medications with an average of 359 medications removed monthly. 428 medications (total quantity of 14,493) were determined to be stagnant medications, unloaded from the ADCs and replaced with more frequently used medications. From the list of stagnant medications, a sample of medication costs were calculated totaling $105,240.

Figure 3.

Change in blind stockout percentage after optimization from baseline. Blind stockout percentage calculated for each automated dispensing cabinet using the equation listed in table 1. The bars indicate standard deviation.

Figure 3.

Change in blind stockout percentage after optimization from baseline. Blind stockout percentage calculated for each automated dispensing cabinet using the equation listed in table 1. The bars indicate standard deviation.

Close modal
Figure 4.

Change in number of medications dispensed per day after optimization from baseline. Medications dispensed per day calculated by adding the number of medications dispensed from each automated dispensing cabinet for each month and averaged by the number of days within that month. The bars indicate standard deviation.

Figure 4.

Change in number of medications dispensed per day after optimization from baseline. Medications dispensed per day calculated by adding the number of medications dispensed from each automated dispensing cabinet for each month and averaged by the number of days within that month. The bars indicate standard deviation.

Close modal
Figure 5.

Panel A shows the vend/fill ratio by month. Panel B shows the stockout percentage by month. Panel C shows the blind stockout percentage by month. Panel D shows the number of medications dispensed per day by month.

Figure 5.

Panel A shows the vend/fill ratio by month. Panel B shows the stockout percentage by month. Panel C shows the blind stockout percentage by month. Panel D shows the number of medications dispensed per day by month.

Close modal
From this study, there was marginal difference seen in the primary performance indicators of vend/fill ratio and stockout percentage when compared with baseline after 10 months of optimization of automated dispensing cabinets. However, there was a significant difference found for blind stockout percentage and number of medications dispensed per day compared with baseline. From these results, we can see that with optimization there was more efficient use of pocket space. The replacement of stagnant medications with more frequently used medications caused a significant increase in the amount of medications being dispensed per day. However, this may have increased the number of refills as well, potentially explaining why there was no significant change in vend/fill ratio based on the equation:
formula

So even though there were more medications being dispensed it was negated by the proportional increase in number of refills. We observed a significant increase in the number of refills compared to baseline strengthening this argument [Difference 30.42 (213.04 ± 157 vs. 183.62 ± 131, respectively; p= 0.0001)].

From the list of stagnant medications, we calculated a sample cost of over $100,000. This number represents the potential cost avoidance associated with the optimization process. Had the optimization process not occurred, these stagnant medications would potentially remain in the pockets until they expired.

While there was not a significant difference found in the primary performance indicators, the optimization process yielded benefit in outcomes that were not measured. During the optimization period, Kaiser Permanente Vacaville Inpatient pharmacy deployed a vial adaptor initiative. The goal of this initiative was to decrease the amount of intravenous (IV) compounding required. Nursing was educated on how to assemble and reconstitute the IV medication themselves by connecting a vial of medication to the appropriate bag of fluid through an add-ease connector (vial adapter). For this initiative to be viable, vials of the IV medication had to be loaded into the ADCs on each floor. With optimization, more pocket space was made available by the unloading of stagnant medications and reorganization of other medications. This allowed for the loading of medications such as the add-ease related IV medications. With this initiative there was a decrease in turnaround time from order to medication administration by ~15 minutes seen in the ED.

Limitations included lack of physical space, a high baseline vend/fill ratio, and relatively small sample size. It was realized early on that physical space limited the optimization process. While there was an increase in useable pockets for more medications, bulkier items such as IV premixes, oral liquids and certain injectables (i.e. enoxaparin) had physical limitations in the cubies. Thus, there was not enough space to expand the inventory of those bulkier items. The lack of towers and only having one main and one auxiliary cabinet per med room at our site prevented the optimization of such larger items as well. Secondly, the vend/fill ratio started off high at 11.43 at baseline. General goal to aim during optimization for the vend/fill ratio is between 10–12. With a high baseline vend/fill ratio, there is less opportunities for improving efficiency of medication utilization. Coupled with the increase in refills that neutralized the gains from the increased medication dispensing, there was little room for improving the vend/fill ratio. Lastly, due to the size of the medical center, the number of ADCs included were lower than expected after applying the exclusion criteria. Compared to other studies mentioned, the sample size of this study is small to moderate size.(2,3)

Overall the optimization process is very feasible requiring only one to two people for four hours once a week to perform the required optimization interventions for a medical center with 40–50 ADCs. This means that pharmacy technicians can easily allocate time once a week or incorporate the optimization process into the regular pharmacy workflow. A caveat is the initial process of unloading and/or loading of medications, and mirroring of ADCs is a time-consuming and laborious process, which may require more than four hours a week to perform in a timely manner.

Potential future improvements to the optimization process identified are the evaluation of other performance indicators with more tangible outcomes and the procurement of towers. Tangible outcomes such as the decrease in medication turnaround time, or nursing and/or pharmacy staff satisfaction may better quantify the benefits of the optimization process. Also, procurement of towers will expand the physical space available in the ADCs removing that limitation at Kaiser Permanente Vacaville.

Optimization of automated dispensing cabinets at Kaiser Permanente Vacaville medical center yielded marginal improvements in vend/fill ratio and stockout percentage and significantly improved overall efficiency through an increase in the number of medications stocked in ADCs and number of medications dispensed per day from ADCs. Evaluation of more clinically significant performance indicators may better characterize the benefits from the optimization process.

1.
Pedersen
CA,
Schneider
PJ,
Scheckelhoff
DJ.
ASHP national survey of pharmacy practice in hospital settings: Dispensing and administration--2014
.
Am J Health Syst Pharm.
2015
Jul
1
;
72
(
13
):
1119
37
.
doi:
.
2.
Lupi
KE,
Day
KM,
Gilmore
JF,
DeGrado
JR.
Evaluation of a Clinical Pharmacist-Led Automated Dispensing Cabinet Stewardship Program at a Tertiary Academic Medical Center
.
J Pharm Pract.
2019
Jan
17
:
897190018823471.
doi:
.
3.
O’Neil
DP,
Miller
A,
Cronin
D,
Hatfield
CJ.
A comparison of automated dispensing cabinet optimization methods
.
Am J Health Syst Pharm.
2016
Jul
1
;
73
(
13
):
975
80
.
doi:
.
4.
McCarthy
BC
Jr,
Ferker
M.
Implementation and optimization of automated dispensing cabinet technology
.
Am J Health Syst Pharm.
2016
Oct
1
;
73
(
19
):
1531
6
.
doi:
.

About the Authors

Tony Vu, PharmD, is an ambulatory care pharmacist currently at Kaiser Permanente Modesto. He graduated from Touro University California College of Pharmacy in 2019 and completed his PGY-1 pharmacy residency at Kaiser Permanente Napa -Solano in 2020. Dr. Vu conducted his research during his PGY-1 pharmacy residency under the mentorship of Dr. Yifan She. Dr. Vu has no conflicts of interest to report.

Yifan She, PharmD, is the inpatient pharmacy director currently at Kaiser Permanente Vacaville Medical Center. He graduated from the University of California School of Pharmacy in 2015. Dr. She is very passionate about optimization of workflows and targeting the goals of achieving operational excellence. Dr. She has no conflicts of interest to report.