2008 ASHP Poster -- Appropriateness of Using a Computer Simulation Approach...

Appropriateness of Using a Computer Simulation Approach in Evaluating the Efficiency of a Units Dose Drug Distribution System

INTRODUCTION
An automated unit dose packaging system, Pyxis Oral Solid Packager (POSP, see Figure 1) system, was put into operation in January, 2008 at the Inpatient Pharmacy of Cincinnati Children’s Hospital Medical Center (CCHMC), Cincinnati, Ohio for filling the most-used oral solid doses. The CCHMC pharmacy management wondered if a computer simulation approach could be used to accurately predict the efficiency outcomes of using the POSP system. This study was conducted to determine the appropriateness of using the computer simulation approach in drug distribution system studies.
Computer simulation is a process of designing a complex model for a real or proposed working system, and it can be a powerful and flexible tool to evaluate the efficiency of a workflow system.

 OBJECTIVE
·         To determine the appropriateness of using the computer simulation approach in reengineering a Unit Dose (UD) picking process
This study compared the efficiency data (UD processing time units and queuing time) obtained from field observations (work sampling) and computer simulation to determine the appropriateness of using the computer simulation approach.

 Operational Definitions
·         Appropriateness of using the computer simulation was evaluated by comparing the variation of efficiency data obtained from field observations (work sampling) and computer simulated models
·         Efficiency Data was defined as the UD process time units (order receiving, order entry, picking, inspection, tubing, and automated UD packaging ) and queuing time.

 METHOD and APPROACH
·         This study applied work sampling and computer simulation techniques. 
·         The study design involved: (1) developing a validated computer simulation model, (2) comparing the efficiency data obtained from the computer simulation approach with a work sampling observation.
·         Prior to the installation of the POSP system, work sampling observation data, showing the time spent by pharmacists and technicians in the categories of “order receiving”, “data entry”, “filling doses”, “inspection”, and “tubing medications” was collected in October, 2007. 
·         The time spent patterns by pharmacy staff from the work measurement observation were used to develop the computer model to simulate the UD filling operations prior to installation of POSP system.
·         Arena 10.0 simulation software published by Rockwell Software, Microsoft Office 2007 was used for developing the simulation model. 
·         After the simulation model was validated, it was modified and used to estimate the efficiency outcomes of using the POSP system.  A work sampling observation was conducted in May, 2008. The performance data between work sampling after installing POSP and computer simulation was compared to determine the appropriateness of using this computer simulation approach.

 Study Site and Automated UD System:
·         UD filling area at Cincinnati Children’s Hospital Medical Center (CCHMC), Cincinnati, OH was the study site. CCHMC is a 423-bed institution and Inpatient Pharmacy opens 24-hour.
·         UD filling area filled an average of 47.12 non-batch UD orders from 7am to 7pm and an average of 275 batch UD orders at around 12 noon. 
·         The average numbers of staff per hour were 2.94 pharmacists and 2.42 technicians.
·         UD doses were delivered by cart, and tube.
·         The POSP system (see Figure 1) was installed to dispense the most-used oral solid doses for the batch UD doses. 

Figure 0. Study Design

 Figure 1.  Pyxis Oral Solid Packaging (POSP) System

 

 Simulation Models
The simulation models require appropriate input data (e.g., order arrival pattern, and processing time units) and logic. The patterns of hourly UD order number and UD processing time was determined by analyzing the computer data and work sampling observation. Those patterns were used in the simulation models to imitate the real-world situation. The logic of simulation models is depicted in Figure 2.
·         Hourly UD order number was determined by using the CCHMC pharmacy computer database. The data was from 7am to 7pm between October 1, 2007 and October 31, 2007. An average of 565.42 orders of accumulated UD order for non-batch UD doses determined and used. The daily batch UD orders ranged between 250 and 300 orders and this information was used in the simulation models. 
·         UD processing time was collected by using work sampling method with one minute fixed-interval observations of the activities of each pharmacy staff. The pilot test was conducted between September 24 and 28, 2007. The pre-installation POSP data collection was conducted from 7am to 7pm between October 1 and 12, 2007 (excluding weekends). The post-installation POSP data collection was conducted from 7am to 7pm between May 5 and 16, 2008 (excluding weekends). The processing unit used in the simulation models is depicted in Table 1.

 Figure 2. Logic of the Simulation Models

 

Table 1. Simulation Input Data: Processing Time

	Observed pre-POSP	Simulated post-POSP	Observed post-POSP
Processing Time per order (in minutes; Mean+S.D.)
Order Receiving*	0.049 (±0.006)	0.049 (±0.006)	0.041 (±0.003)
Order Entry*	2.736 (±0.472)	2.736 (±0.472)	2.917 (±0.309)
Fill Non-Batch Orders*	1.772 (±0.195)	1.772 (±0.195)	1.506 (±0.132)
Fill Batch Orders*	1.16 (±0.159)	1.089 (±0.121)	1.089 (±0.121)
Inspection*	0.735 (±0.149)	0.735 (±0.149)	1.022 (±0.098)
Tubing  Medication*	0.237 (±0.039)	0.237 (±0.039)	0.286 (±0.029)
POSP Processing time**		Min: 0.0938 Max: 0.125	Min: 0.0938 Max: 0.125

*Normal Distribution; **Uniform Distribution

 Analysis of the Appropriateness of Using Computer Simulation Approach
 Two internal validations and one external validation were conducted to examine the appropriateness of using the computer simulation approach in reengineering the UD picking process.
·            Pre-POSP and post-POSP model internal validation: compare observed data and simulation results. Internal validation is to validate the logic of the simulation models.  (Figure 3)
·            External Validation: compare post-phase simulation result and predicted post-phase simulation result. External validation is to validate the appropriateness of using computer simulation to predict a changed system. (Figure 3) 

RESULTS
·         There was no difference between observed data and simulated result in the order processing time:
w  Using the pre-POSP data. (see Table 2: Non-batch orders: 5.528 min versus 5.571 min; Batch orders: 1.895 min versus 1.894 min). 
w  Using the post-POSP data. (see Table 2: Non-batch orders: 5.773 min versus 5.573 min; Batch orders: 2.111 min versus 1.928 min).
·         For the queuing time analysis of non-batch orders, there was no significant difference between simulated post-POSP versus post-POSP observed data and pre-POSP versus post-POSP observed data, but there was a significant difference between these two groups after POSP was installed. (Table 3)
·         For the queuing time analysis and processing time of the batch orders, there were both significant differences between simulated post-phase versus post-POSP observed data and pre-POSP phase versus post-POSP observed data after using POSP system. (Table 3)

Table 2.  Internal Validation of Simulation Model

 

	Unit: in Mintes
	Non-batch Order (Simulation)	Non-batch Order (Observation)	Batch Order (Simulation)	Batch Order (Observation)
Pre-POPS Internal Validation
Process time per order	5.571	5.528	1.894	1.895
Queuing time per order	13.717	N/A	13.799	N/A
Post-POSP Internal Validation
Process time per order	5.573	5.773	1.928