Office of Clinical Affairs: Evaluation of the Medication Manager Pilot s Effects at Mott Children s Hospital. University of Michigan Health System

Office of Clinical Affairs: Evaluation of the Medication Manager Pilot s Effects at Mott Children s Hospital University of Michigan Health System Final Report April 21, 2008 Client: Deb Wagner, Pharm D Pediatric Safety Coordinator Coordinator: Jackie Lapinski Project Manager, Children s and Women s Project The Program & Operations Analysis Department Michael Bagliebter, IOE 481 Student Brandon Barnett, IOE 481 Student Matt Cooper, IOE 481 Student Julian Stauffer, IOE 481 Student

TABLE OF CONTENTS Executive Summary i Background i Methods and Findings i Nursing and Pharmacy Surveys i Nursing Time ii Missing Medication Costs ii Summary and Recommendations iii Introduction 1 Background 1 Goals and Objectives 2 Expected Impact 3 Project Scope 3 Key Issues 4 Methods and Findings 4 Nursing and Pharmacy Surveys 4 Method: Surveys 4 Finding: Nursing Surveys 5 Findings: Pharmacy Surveys 6 Nursing Medication Retrieval Time 8 Method: Nurse Trip Tally Sheets 8 Method: Nurse Travel Time Studies 9 Method: Omnicell, CareLink, and Pager Logs 10 Findings: Nursing Medication Retrieval Time 10 Missing Medication Cost 12 Method: Discarded Medication Count 12 Method: Medication Manager Logs 12 Method: Pharmacy Phone Logs 12 Method: CareLink Data 13 Method: Medications Dispensed and Cost Data 13 Findings: Missing Medication Costs 13 Other Notes on Validation 16 Summary 16 Nursing and Pharmacy Surveys 17 Nursing Medication Retrieval Time 17 Missing Medication Costs 17 Recommendations 17 a

TABLE OF CONTENTS FOR TABLES AND FIGURES Figure 1. Layout of PICU and PCTU 1 Table 1. Nursing surveys indicate appeal of Medication Manager 5 Figure 2. Nurses responding 4 or 5 increased from ~40% to 95% with Med. Manager 5 Figure 3. 100% nurses in 2008 study indicate 4 or 5 for knowing medication location 6 Table 2. Pharmacy responses mostly average above 4 7 Figure 4. Nearly identical response distribution as 2007 study 7 Figure 5. Pharmacy employees continue to know where to deliver medications 8 Table 3. Time study recording standards 9 Figure 6. Effective nursing non-value added time costs, showing nearly a 50% reduction 11 Table 4. Summary of nursing time study data showing large decreases in med. room values 11 Table 5. Higher cost per wasted medication and higher annual wasted medications in 2007 13 Figure 7. ~$110K savings in missing medication costs based on 2007 methodology 14 Table 6. Report listings of PCTU and PICU dispensed doses and costs 14 Table 7. Daily medications and costs 15 Figure 8. ~$42,000 savings with medication manager 15 Table 8. Average occupancy slightly higher in 2008 study 16 b

Executive Summary Background The C.S. Mott Children s Hospital is known worldwide for its excellent clinical, educational and research programs. As a leader in pediatric care, the hospital must maintain a high standard of healthcare. As a small step forward in patient care, the Pediatric Intensive Care Unit (PICU) and Pediatric Cardio-Thoracic Intensive Care Unit (PCTU) have implemented a Medication Manager Pilot. The Pilot is aimed at increasing nursing time at the patients bedsides, while simultaneously making medicine more available to nurses. In the pre-pilot system, nurses traveled to the Medication Rooms or the Pharmacy to pick up medications for their patients. This practice wasted valuable patient care time and potentially led to missing medications. In the pilot program, three pharmacy technicians act as medication delivery personnel. They are responsible for delivering the patients medications to their respective bedside drawers. This reduces nursing time spent away from bedsides. While most medications are placed at the patients bedsides, nurses are still required to retrieve narcotics from their unit s Omnicell, which is a secured medication dispensing machine. The hospital asked the Bagliebter, Barnett, Cooper, and Stauffer team (BBCS) to quantify both the decrease in wasted nursing time spent away from patients bedsides due to medication retrieval and the hospital s monetary savings due to the decrease in missing medications. The base metric for all measurements was derived from the study performed by DiMusto et. al. in the Fall of 2007. In this report, BBCS will present its methodologies, findings, and recommendations for the Medication Manager Pilot in the PICU and PCTU by quantifying the cost of nursing time to retrieve medications and the cost of missing medications. Methods and Findings To quantify the cost and impact of the Medication Manager Pilot, BBCS collected various pieces of data from the two units and the pharmacy. The team focused its evaluation of the Medication Manager Pilot on the nursing time allocated to medication retrieval, the cost of missing medications, and nursing and pharmacy perception of the program. The following methodologies and findings are a result of the team s studies. Nursing and Pharmacy Surveys Surveys were distributed to the nursing and pharmacy staff to examine their perception of the Medication Manager Pilot. The team used the surveys to supplement and quantify informal nurse and pharmacy interviews. Review of the surveys revealed that 95% of nurses usually or always know when their medications are delivered from the pharmacy. This is an increase from 40% in the Fall 2007 study. Also, 100% of the nurses usually or always know where their medications are delivered. This is a large improvement from the Fall 2007 report where 0% always knew where their medications were delivered. The pharmacy surveys showed little difference from the pre-pilot study, which is an indication that there has been little effect on pharmacy operations due to the implementation of the i

Medication Manager. The pharmacy surveys and the nursing surveys both showed a strong perception that the medication manager improves patient care. The pharmacy surveys and the nursing surveys averaged an over 4 out of 5 (agree to strongly agree) response when asked if the Medication Manager improves patient care. Nursing Time The BBCS team performed various studies in the PICU and PCTU to determine the amount of nursing time spent away from the patients bedsides due to the implementation of the Medication Manager. The findings of these studies were compared to the findings of the Fall 2007 study to show any changes in nursing time. The team completed a total of 25 hours of time studies in the PICU and PCTU. The data collected shows that the nursing walking time and walking routes are consistent with the previous group s data. A significant decrease was found in the amount of time that nurses spend in their respective Medication Rooms. This finding can be attributed to the availability of medications at the bedside under the new system. To validate the data found, Omnicell data, CareLink information, and pager log data were pulled from the hospital. This data was used in an equation developed by BBCS to determine the overall time nurses spend away from the patients bedsides. In addition to the time study, an auxiliary study was performed to obtain a metric for the frequency that nurses look for missing medications, and the channels that they use to obtain the medications. These channels include asking the Medication Manager, walking to the Medication Room, walking to the Omnicell, or walking to the pharmacy. Nurse trip tally sheets were distributed to the nurses in both units for a three week period. Nurses were asked to mark the sheets each time they departed the bedside in search for patient medications. Once combined, the nursing time per trip away from the bedside and the frequency of the trips indicated a reduction in non-value add time retrieving medications of about 50% with the implementation of the Medication Manager Pilot. The team calculated a savings of $16,500 annually in equivalent salary (from $35,600 to $19,100) with the Pilot. It is important to note that this cannot be realized as economic savings. The team found that the time per bed per day spent retrieving medications was reduced from about 8 minutes to 4 minutes. Missing Medication Costs The BBCS team performed a two-week daily count of any expired or unusable medications collected by the pharmacy to estimate the value of wasted medications. Each disposed medication was matched with its corresponding cost based on its type, dose, and route. The costs of individual medications were provided by the pharmacy staff, which BBCS used to calculate wasted medication costs. Missing medications as a percentage of wasted medications was then determined and multiplied to give a value of missing medications. The cost of missing medication is an important metric for the study of the Medication Manager Pilot. The number of missing medications was determined by asking the Medication Managers to keep track of medication requests due to missing medications over the two week period. Based on the data collected by the Medication Managers, there is an average of 5.8 missing medication requests per day, versus 40 per day before the Pilot. The data collected from the Medication Managers was used in conjunction with CareLink data to help determine the ii

frequency with which medications were missing. Using this determined number of missing medications, the cost of missing medications was calculated to be $11,000 annually. This is a reduction of 112,000 in missing medications when compared to the results of the Fall 2007 Study. Although this is a significant drop in missing medications, there is a high potential of variation with this costing method. Because of this variation, the BBCS team examined the annual cost of missing medications using a second method, which involves the total cost and quantity of medications dispensed. The hospital provided BBCS with the quantity and cost of dispensed medications in the PICU and PCTU for specific time periods. The percentage of dispensed medications that were missing was determined by using the quantity of missing medications and the total number of dispensed medications over the two week period. Using this information BBCS determined that 0.6% of them are considered missing. In the Fall 2007 report, 4% of all dispensed medications were considered missing. The cost of dispensed medications that were missing was calculated by multiplying the percentage of missing medications by the total cost of all dispensed medications. Therefore, the cost of missing medications was estimated to be $7000 annually, which is $42,000 less than the cost calculated for Pre-Pilot conditions. The team believes this is a more accurate representation Summary and Recommendations The team s findings indicate that the nursing staff considers the Medication Manager Pilot to be helpful. This notion is supported in the findings, which indicate a reduction in medication retrieval time of 50% and a reduction in missing medication costs of about 85%. The reduction in missing medications may also impact the time spent making medications in the Pharmacy. However, it must be noted that the potential for savings from reducing missing medications is much less than originally determined in the Fall 2007 Study. When compared to the Medication Manager Pilot costs, estimated to be about $100,000 per year, the economics are not as optimistic. The team recommends standardizing the Medication Manager System. These changes would include improving medication request records, reducing channels for medication requests, and placing more emphasis on the Medication Administration Record. Once the changes are implemented and results are determined, the costs and benefits of the Medication Manager may be reassessed. iii

Introduction The University of Michigan Hospital System (UMHS) is committed to improving patient care. Therefore, the hospital is interested in improving the continuity of care to patients by increasing the time nurses directly care for patients. Additionally, UMHS must balance improved patient care with an economic responsibility to keep costs down. To reach this dual objective, the Office of Clinical Affairs has implemented a Medication Manager Pilot in the PICU and PCTU of Mott Children s Hospital. This Pilot utilizes a dedicated Medication Manager who retrieves medications from the pharmacy and places the medications into a locked cabinet at patient bedsides. The pilot aims to reduce time nurses spend retrieving medication and decrease costs from missing medications. The University of Michigan Health System has requested quantitative missing medication and nursing time data to analyze the effectiveness of the Medication Manager Pilot. BBCS quantified the results of this pilot and determined the viability of the Medication Manager Pilot. Data from a 2007 pre-pilot study was used to create a baseline system state to quantify any improvements. The team determined that the Medication Manager leads to a reduction in medication retrieval time for nurses of 50% and a reduction in missing medication costs of 85%. However, the cost reduction, about $42,000, is a lower amount than anticipated before the study. This report contains the team s methods and findings obtained throughout the duration of this study. Background The Medication Manager Pilot was launched on January 21, 2008 in the PCTU (Pod A) and PICU (Pods B and C) of Mott Children s Hospital (See Figure 1). Figure 1. Layout of PICU and PCTU Pod A Pod B Pod C Pharmacy Medication Rooms 1

Prior to the implementation of the Medication Manager Pilot, the nurses spent excess time retrieving patient medications from Medication Rooms or Omnicell dispensing machines. This system led to numerous problems, analyzed in the Fall 2007 study by Healthcare Strategies and Solutions (HSS), Evaluating the Current State of Medication Retrieval at Mott Children s Hospital 1. (Note that the current state referred to in the study s title is no longer current, since the status quo in Fall 2007 has been replaced by the Medication Manager Pilot.) This study identified two critical problems with the system. First, nursing time with patients was reduced because nurses were required to retrieve medications from distant locations and bring them to the patient. The value of nursing time spent on medication retrieval for the PICU and PCTU was deemed to be approximately $36,000 2. Second, communication between the nurses and the Pharmacy was considered substandard. The key metric that led to this conclusion was medication wasted due to lost or missing medication. Losing medications often occurred because inefficient communication channels led to the medication not arriving at the proper location at the proper time. Loss was estimated at $121,000 annually. The Pediatric Safety Coordinator has also directly informed BBCS of her concerns about the prepilot medication distribution methods. According to the Pediatric Safety Coordinator, the delivery of medications to patients could have been delayed. Another issue with the pre-pilot process was the cluttered space of the Medication Room. Several nurses working in and around the Medication Room created confusion, which possibly led to medication retrieval issues. The Medication Manager Pilot tries to remedy these problems by shifting the responsibility of medication distribution to the Medication Managers. Therefore, orders placed by the nurses with the pharmacy are delivered to the patients rooms by these dedicated Medication Managers. This means that the Medication Managers hand the medications directly to the nurses or they place the medications into secure drawers at each bedside. The Medication Managers have a cell phone and pager with them for medication requests. The nurses can also request medication from the Medication Managers directly. Goals and Objectives The primary objective of this study was to determine the extent to which the Medication Manager Pilot improves upon the previous system. Additionally, the team determined whether these improvements led to a cost savings for the hospital. To determine the improvements made by the Pilot, the team performed the following tasks: Quantified the amount of time nurses spend locating and retrieving medications Quantified waste due to missing medications Interviewed nurses, pharmacists, and Medication Managers to determine ways to improve the Medical Manager Pilot, as well as to determine their views about the Pilot. 1 DiMusto et al, 2007 2 Actual estimate: $35,600 2

Comparing these findings to the findings in the 2007 Study enabled the team to achieve its primary objective. Expected Impact Based on extensive data collection and calculations using medicine and nurse salary costs, the final outcomes of the project include: Economic savings in minimizing missing medication Effective salary savings due to reduced time in nursing retrieval of medication Determination of whether increased patient care and reduced medication waste justifies the cost of the Medication Manager General impressions from nursing and pharmacy staff Recommendation for improvement of Medication Manager program Recommendation for future Medication Manager Pilot status The project will allow the hospital to make informed decisions concerning a permanent installation of the Medication Manager in the PICU/PCTU, as well as an extension of the Medication Manager to other units in the hospital. Project Scope This project analyzed the current state of the pilot s medication delivery process in the PICU and PCTU at the Mott Children s Hospital. To determine the improvement from the pre-pilot analysis, the BBCS team collected the following data: Number of calls, pages, and in-person medication requests made to Medication Managers Number of nursing calls to the pharmacy to order medications Number of nursing orders via CareLink for medications Number of orders through all channels that are due to missing medications Number and type of disposed medications Number of trips nurses make to the Med Room Number of trips nurses make to the Omnicell Number of trips nurses make to the Pharmacy Number of missing medications requests from the medication logs and CareLink Time a nurse takes to retrieve medication (travel and handling) This project did not involve any raw data from before the onset of the pilot, although this specification will not limit the use of pre-pilot statistics from the Fall 2007 Study for improvement analysis. Also, the project did not address errors in dosages for medications. Additionally, the project did not include data collection on the night shifts, which means only the 3

7 AM and 3 PM shifts were covered. By excluding these activities, the team was able to offer a more focused and precise analysis of the Medical Manager Pilot program. Key Issues To maintain consistency and create a valid baseline for comparison, the BBCS team used nearly the same methodologies from the 2007 project by HSS. There is one change and one addition to the methodologies that are worth noting because they increased the accuracy of the findings. For the time studies, BBCS determined how much time the nurses spend either in the Medication Room or at the Omnicell. The time to travel between the bed and Medication Room or Omnicell was not measured, as it was determined in the Fall 2007 Study, and was assumed to be constant. The BBCS team also surveyed the Medication Manager, due to the addition of the staffing position, which did not exist in the previous study. Methods and Findings The BBCS team collected and analyzed data in three distinct categories: Nursing and pharmacy surveys Nurse medication retrieval time Missing medication costs Nursing and Pharmacy Surveys To collect sentiment in regards to the Medication Manager Pilot, BBCS distributed unit-specific, five-question surveys on March 24 th, March 25 th and April 9, 2008 to the nurses and Pharmacy employees. The surveys consisted of two parts. Part one on both surveys consisted of two questions consistent with the Fall 2007 Study. The questions for part two claimed three different statements asking the participant for a level of agreement. All questions on the survey were related to patient care improvement and Medication Manager Program improvement. Method: Surveys To survey the nurses and the Pharmacy employees, BBCS visited the PICU and the PCTU at 7:00pm on March 24, 10:00 am on March 25, and again on April 9, 2008. These three visits allowed BBCS to get one afternoon shift and two morning shifts. BBCS distributed the surveys throughout the units during these shifts and collected them when completed. The survey questions were scaled from 1 to 5 with 1 being Strongly Disagree or Never and 5 being Strongly Agree or Always. The survey was divided into two parts. Part one consisted of the questions for comparison with the Fall 2007 Study, while part two were questions designed by BBCS to quantify nursing sentiment about the Medication Manager Program. The survey also provided space for further comments, where nurses openly expressed their thoughts about the Medication Manager Program. The nursing and pharmacy surveys as well as nurses comments are located in Appendices A, B, and C respectively. 4

Findings: Nursing Surveys In the PICU and the PCTU thirty-three nurses were surveyed. The table below states the questions asked, as well as the responses received and average responses quantified. Table 1. Nursing surveys indicate appeal of Medication Manager Nursing Surveys, n = 33 Question 1 2 3 4 5 Avg Part 1 Never Always How often do you know when your medication arrives from the pharmacy? 0 0 2 23 8 4.18 How often do you medications get delivered to where you expect? 0 0 0 10 23 4.70 Part 2 Level of Agreement The Medication Manager improves patient care. 0 0 0 4 29 4.88 I always know where my medications are located 0 1 0 10 22 4.61 There are improvements that can be made to the new medication retrieval system. 6 8 14 3 0 2.45 N = 33, March 2008, by BBCS Figure 2 and Figure 3 show the comparison between the Fall 2007 Study (pre-pilot) and the current state of the system. Figure 2: Nurses responding 4 or 5 increased from ~40% to ~95% with Medication Manager 2007: N = 57, mean = 2.4, October 2007, by HSS 2008: N = 33, mean = 4.2, March 2008, by BBCS 5

Figure 3: 100% nurses in 2008 study indicate 4 or 5 for knowing medication location 2007: N = 57, mean = 3.6, October 2007, by HSS 2008: N = 33, mean = 4.7, March 2008, by BBCS Both figures indicate a significant increase in nurse knowledge of the location of medications upon delivery. Figure 1 shows that nearly 95% of nurses usually or always know when their medications are delivered from the Pharmacy. This is nearly a 55% change from the Fall 2007 Study. Figure 2 indicates a similar shift as Figure 1. Under the current state of the system 100% of the nurses usually or always know where their medications are delivered. These changes show the impact the nurses feel the Medication Manager has had. Part two of the survey consisted of two questions which are worth discussing. In this part of the survey a statement was made and the nurses were asked their level of agreement with the statement. Here 100% of the nurses believed that the Medication Manager improves patient care. The nurses were also questioned if improvements should be made to the new medication retrieval system. In response most nurses were either neutral to the question of felt that no changes were needed. Written comments provided by nurses showed that there is a very positive response to the Medication Manager. The most frequent comments involved the decreased wait time for STAT medications, increased bedside care time, and increased overall medication availability due to the program. Nurses commonly say that the Medication Manager is extremely helpful, makes their lives far easier and should absolutely continue as a program. There were a few concerns, however. A nurse said that delivery can sometimes be repetitive and wasteful when doses change, and another noted a medication request program glitch. Detailed survey response data is available in appendix C. Findings: Pharmacy Surveys A combination of ten technicians, pharmacists, and medication managers were surveyed at the Satellite Pharmacy on the 5 th floor of Mott. The results of the survey showed little effect due to 6

the implementation of the Medication Manager. The table below states the questions asked, as well as the responses received and average responses quantified. Table 2. Pharmacy responses mostly average above 4 Pharmacy Surveys, n = 10 Question 1 2 3 4 5 Avg Part 1 Never Always How often do you feel a missing order has previously been processed and dispensed? 0 0 1 6 2 4.11 How often do you know where within a unit to deliver the medications? 0 0 0 2 7 4.78 Part 2 Level of Agreement The Medication Manager improves patient care. 0 0 3 1 6 4.30 I always know where to deliver medications. 0 1 0 5 3 4.11 There are improvements that can be made to the Medication Manager's tasks. 0 1 7 2 0 3.10 The figures 4 and 5 below show a comparison between the current state and the pre-pilot state. Figure 4: Nearly identical response distribution as 2007 Study 7

Figure 5: Pharmacy employees continue to know where to deliver medications As shown in the charts, there is little variation between the current state and the pre medication manager state. Therefore this is an indication that those employed in the Pharmacy do not feel the effects of the addition of the Medication Manager. Nursing Medication Retrieval Time To determine the change in nursing time spent away from patients bedsides due to the implementation of the Medication Manager, BBCS performed several different studies. Nurse trip tally sheets were distributed, which were to be marked each time the nurse departed a patients bedside while searching for a medication. A nursing time study was completed, which was aimed at determining the total time nurses spend at the Omnicell and in the Medication room due to the implementation of the Medication Manager Pilot. Various data pieces were then pulled from hospital databases. Using all of this data, BBCS formed an equation which included the nurse trip totals, time study information, Omnicell data, CareLink information and pager logs to determine nursing time spent away from the patients bedsides. Method: Nurse Trip Tally Sheets The nurse trip tally sheets distributed during the Fall 2007 Study were altered for this Medication Manager Pilot study. The purpose of the tally sheet was to quantify the number of trips nurses completed when looking for medications around their respective units. Nurses were requested to mark the location in which they searched for their patients medications and whether they found the medications. The locations on the sheets included the Medication Room, the Omnicell, and the Pharmacy. Another column on the tally sheet was provided for requested medications by means of the Medication Manager. The nurses in both units completed the tally sheets from March 6, 2008 to March 26, 2008. Each trip tally sheet covered a twenty-four hour period. Every midnight from March 6, 2008 to March 26, 2008, the clerks in the PCTU and PICU placed the nurse trip tally sheets on the flow boards of each bed. Each separate line on the tally sheet represented a search for a single medication, regardless of the number of locations visited. On 8

March 5, 2008 and March 6, 2008, BBCS attended all of the nursing shift change meetings to inform the nurses of the tally sheets and field questions regarding the sheets. The nurses completed the tally sheets on a daily basis and BBCS collected these sheets on a daily basis. Of note, 335 tally sheets were recorded, for a completion rate of approximately 57%. The team deemed this number of sheets to be fully acceptable for accurate results. Method: Nurse Travel Time Studies A time study was performed to quantify the time that nurses spend inside the Medication Rooms and at the Omnicells. BBCS collected 25 total hours of time study data. To ensure accurate data records, the BBCS team used the same standards as used in the 2007 Study, as seen in Table 3. Table 3. Time study recording standards Time Measured Start Point End Point Walk Time Bedside Nurse steps out of the patients room Medication Room Nurse leaves the Medication Room Nurse steps into the patients room Nurse touches the key pad to gain access into the Medication room Nurse steps up to the Pharmacy window Pharmacy Nurse steps away from the Pharmacy Omnicell Omnicell logs out the Nurse Nurse touches the Omnicell key pad Time in Medication Room Medication Room Only Medication Room from Omnicell Nurse pushes the first button on the key pad to gain access into the medication room Omnicell logs out the nurse inside the medication room Time at Omnicell Nurse first pushes the button on the Omnicell Time at Nurse steps up to the Pharmacy Pharmacy window Nurse s foot crosses the threshold of the Medication room Nurse first pushed a button on the Omnicell in the Medication room Omnicell logs out the nurse Nurse takes his/her first step to walk back to their respective pod In order to collect data on the Medication Room and the Omnicell task durations, the BBCS team members stationed themselves adjacent to theses locations. BBCS timed the nurses while they obtained medications from either location. In several instances, the Omnicell was located within the Medication Room. BBCS asked the nurses to specify if they visited the Omnicell during their time within the Medication Room. This data was collected during in-scope nursing shifts. 9

The team opted, in the middle of data collection, to cease recording time spent walking to and from locations. The team assumed that the times for these events would be approximately equivalent to the times found in the 2007 Study. Therefore, the results from the 2007 study were used. Method: Omnicell, CareLink, and Pager Logs The BBCS team obtained data from the Omnicell database from the Senior Applications System Analyst/Programmer. This information includes how many medications were removed or delivered as well as the log in and log out times at each Omnicell. Additionally, CareLink data and Pager Logs were used to aid in accuracy of the nurse trip tally sheets. Findings: Nursing Medication Retrieval Time In performing the nursing time study, walking times were collected to verify that the nursing routes were consistent with the Fall 2007 study. Average walking times were found to be statistically similar to the previous report s data, and so this aspect of the time study has been omitted from this report. The team performed a quick validation of Omnicell data. The team found average time per login to the Omnicell to be 54 seconds. Data pulled from the Omnicell database indicated an approximate average of 47 seconds per login. This difference was not considered critical to results, and 54 seconds was used in calculations. Additionally, the number of trips per day to the Omnicell was estimated. The team determined that there were approximately 77 trips per day. In the Omnicell data, the activity of 22 nurses yielded 541 trips over 21 days, or 25.8 trips per day. As a rough calculation, if we assume about three times that number of nurses in the PICU and PCTU (2 per bed), we have 3 * 25.8 = 77.4 trips per day. This number is in line with our original estimate. The nurse trip tally sheets that were distributed and filled out by nurses were an important characteristic of the nursing time studies. In order to better estimate the actual amount of trips that nurses make, the team created an equation to rule out any noncompliance bias. The proportion of CareLink requests that the nurses marked on the sheets versus the number that the pharmacy recorded was inverted and then multiplied by the number of medication room visits. This extrapolation should remove any bias, and create a far better estimate of medication room visits. Based on a combination of nursing time data and trip data, BBCS determined that the Medication Manager Pilot creates annual effective savings of approximately $16,000 in nursing time. The data are summarized in Figure 6 below. It is important to note that while the nursing time may quantify to this 50% reduction in non-value add nursing salary costs, it is an unrealized cost. This is because there is no reduction in nursing salaries. Rather nurses are just using their time more effectively at patients bedsides rather than searching for medications. It is also unreasonable to assume that the nurses will shift all non-value add time into purely value add or effective time, therefore even these effective savings may be slightly out of proportion. 10

Figure 6. Effective nursing non-value added time costs, showing nearly a 50% reduction BBCS also looked at the time savings that the Medication Manager Pilot offers nurses, as a metric for improvement in patient care. Using the trip tally sheets, BBCS found that the two most significant trip reduction values are the average time that nurses spend in the Medication Room, as well as the average trips per day to the Medication Room. Both reductions are found to be a direct result of the Medication Manager Pilot. These values are summarized below in table 4. With the addition of the Medication Manager, nurses make far less trips to the Medication Room because the medications for their patients are now, more often, available next to the bedside in a locked drawer. Also, the reduction in time spent at the Medication Room can be attributed to the nurses no longer collecting medications that their patients would potentially need in the future in order to save extra trips. Nurses notified BBCS of this method during informal interviews in both units. Nurses often used this method to eliminate excess trips to the Medication Room. Table 4. Summary of nursing time study data, showing large decreases in Medication Room values 2007 Study 2008 Study Avg. time at Omnicell (sec) 54 54 Avg. time in Med Room (sec) 43 25 Trips / day, Omnicell 96 77 Trips / day, Med Room 206 30 As a result of the decrease in average Medication Room visits and decrease in frequency of the visits, BBCS determined that nurses save time due to the reduction of medication trips. However, the reduction in trips only contributes a time savings of approximately four minutes per bed per day. 11

Missing Medication Cost The critical economic component of the study focused on the potential cost savings the Medication Manager could bring about in the PICU and PCTU by reducing missing medications. Method: Discarded Medication Count The BBCS team visited the pharmacy to count medications that were disposed. These medications were placed in a designated disposal bin to ensure the BBCS team could locate the medications. Counting of medications began on March 6, 2008, and was completed on March 19, 2008. The process included indicating the type, dose, and route (e.g. intravenous, oral) of the discarded medications. Upon completion of the counting, costs for each medication were matched with the disposed medications to determine medication costs. One of the key metrics for the Medication Manager Pilot is the cost of missing medication due to miscommunication between the pharmacy and nurses. Missing medication can be defined as medication that goes to waste because it cannot be located at the necessary time. It is hoped that the Medication Manager is able to reduce the amount, and thereby the costs, of missing medications. For a fourteen day period, the BBCS team tracked wasted (discarded) medications for the PICU and PCTU. Each day, a team member would go to the pharmacy and indicate the type, dosage, and route of administration for each wasted medication. The wasted medications had been placed in a container specifically for BBCS to count. The data for each of the fourteen days were combined at the end of the data collection. Once combined, costs for each wasted medication were determined. A cost database indicated a cost per unit of medication. Multiplying this cost per unit by the actual amount of medication disposed of yielded cost per medication. For instance, if Medication X cost $0.40/mg, and 10 mg were dispensed, the cost for that medication would be $4.00. These were summed to give an indication of total wasted medications for the fourteen day period. Method: Medication Manager Logs Medication Managers record medication orders from nurses on a log sheet. BBCS arranged to have the Medication Managers indicate whether the order was due to a refill or a missing medication. The data collected from these logs was used in conjunction with CareLink data to help determine the frequency with which medications were missing. Method: Pharmacy Phone Logs The BBCS team distributed phone logs to the pharmacy to quantify the number of missing medication occurrences reported by the nursing staff. The logs were first distributed on March 6. For every medication order the pharmacy received via phone, the unit (PICU, PCTU, or other) that made the call was indicated. Additionally, the pharmacist receiving the call marks whether the medication is requested because a previous dose was missing or not. Several days after distribution of these forms, they were recalled. Under the Medication Manager Pilot, nurses no 12

longer call the pharmacy for medications, but rather talk directly with the Medication Manager. Phone log responses were deemed to be insignificant to the team s findings. Method: CareLink Data BBCS received CareLink data from March 6, 2008, to March 19, 2008, a fourteen day period. The medication requests on CareLink indicated whether a medication request was due to a refill or a missing medication. The number of requests for each reason was determined. Method: Medications Dispensed and Cost Data For the 2008 study, missing medication costs were determined a second way. Rather than using missing medications as a percentage of wasted medications and using wasted medication costs, the team used data pulled from the Pharmacy system. Cost per missing medication could be determined, and the number of missing medications per year could be determined. This led to calculations for the cost of missing medications per year. Also, missing medications as a percentage of total medications dispensed was calculated. Findings: Missing Medication Costs The team first calculated the wasted medications using the methodology from the 2007 Study. The team compiled the information from the 14 days of collecting the type, dose, and route of all wasted medications and then determined the cost of each medication (Table 5). These figures were compared with those from the 2007 study. Table 5. Higher cost per wasted medication and higher annual cost of wasted medications in 2007 Study 2007 Study 2008 Study Total Number, Wasted Meds 1663 1736 Total Cost, Wasted Meds $13,985 $8,659 Number / Day, Wasted Meds 119 124 Cost / Day, Wasted Meds $999 $618 Cost / Wasted Med $8.41 $4.99 Annualized Cost, Wasted Meds $364,608 $226,752 In this table, one of the reasons to be skeptical of the 2007 methodology becomes apparent. In 2007, the cost per wasted medication is nearly $3.50 higher than in 2008. The team expected that cost per wasted medication would remain approximately constant. The team believes that the high variability in wasted medications is the cause of this difference. In fact, analyzing data from the 2007 study shows that just 12 medications accounted for 68% of the total cost of wasted medication. These high-cost medications (some costing greater than $2,000) were not seen in the 2008 study, leading to the lower cost per wasted medication. Next, the team calculated the second critical input, missing medications as a percentage of total wasted medications. The team recorded 34 missing medication requests on CareLink and 47 missing medication requests via the Medication Manager, totaling 81 missing medication requests in the 14 day period. The team calculated the percentage of total wasted medications due to missing medications as (81 requests / 1736 medications) *100% = 4.67%. The team then multiplied by the annualized cost of wasted medications to get an annual cost of missing 13

medications equal to (4.67% * $226,752) = $10,589. We compared these findings with those from the 2007 study in Figure 7. Please not that 2007 study results were recalculated. Figure 7. ~$110K savings in missing medication costs based on 2007 methodology 400 Missing and Non Missing Medications 300 123 $ '000s 200 100 11 242 215 Missing Non Missing 0 2007 2008 Study Figure 7 indicates approximately $110,000 in annual gross savings due to the reduction in missing medication costs because of the Medication Manager pilot. However, the team believes that the updated methodology will provide a better indication of true missing medication costs per year. We received the data in Table 6 from a Pharmacy database in the hospital. Table 6. Report listing of PCTU and PICU dispensed doses and costs Facility Code Area Timeframe Tot Disp Unit Qty Tot Unit Cost MCHC PCTU Mar 5 Mar 28, 11,455 $16,309 2008 MCHC PCTU Oct 22 Nov 22, 2007 18,482 $39,337 MCHC PICU Mar 5 Mar 28, 13,167 $74,839 2008 MCHC PICU Oct 22 Nov 22, 2007 13,981 $62,950 Though ideally the team would have received data for the exact periods that the report was conducted and a full annual count of dispensed medications and costs, the team deemed the data sufficient for accuracy. The team calculated dispensed medications per day, cost per medication, and other information from this data, seen below in Table 7. 14

Table 7. Daily medications and costs 2007 2008 Medications Dispensed / Day 1014 1026 Cost / Day $3,196 $3,798 The aggregate average cost per medication dispensed from the two periods was $3.39. This figure was multiplied by the missing medication requests recorded for each study and then annualized to determine the cost of missing medications per year. Additionally, missing medication requests as a percentage of total medications dispensed was calculated. Figure 8. ~$42,000 savings with Medication Manager 100 Missing and Non Missing Medications $ '000s 49 0 2007 2008 Study 7 The second methodology indicates an annual cost of $49,000 based on the 2007 study and $7,000 based on the 2008 study for missing medication costs. This represents a savings of ~$42,000 in missing medication costs due to the implementation of the Medication Manager. Additionally, the team found that pre-pilot, 4.0% of all medication requests were from missing medications. After implementation of the Medication Manager, this figure dropped to 0.6%, a reduction of approximately 85%. The two methods of calculating annual missing medication results yielded savings numbers differing by about $70,000. Though both methods appear at first valid, the second method appears to be the most accurate. There is a much larger sample of data (about 55,000 medications instead of about 3,000). Also, upon reflection, using wasted medications en route to determining missing medications requires using data much more dependent on day-to-day performance and therefore has higher variability. To demonstrate, if we assume equal missing medication percentages for both pre-pilot and during the Medication Manager Pilot, the data recorded in 2007 would still have much higher total wasted medication costs. Percentages for missing medications of all medications dispensed appears to be a more intuitive number than missing medications as a percentage of waste (which should also be reduced.) 15

A key metric is the actual economic impact of the Medication Manager. The team calculated the gross savings in reduced missing medications from the Medication Manager to be between $40,000 and $110,000. However, the true value is expected to be closer to $40,000. The cost of the Medication Manager includes salaries, benefits, and training. There are 2.8 FTE (full-time equivalent) pharmacy technicians with estimated annual salaries of $28,000 plus 31% benefits. Training costs are unknown. Though perhaps slightly higher in actuality (~$105,000), the original estimation of $100,000 for the cost of the Medication Manager is relatively accurate. A comparison of the costs and savings leads to a range of approximately -$60,000 to $10,000 for net savings in purely economic terms due to the Medication Manager. This figure is probably closer to the negative side of the spectrum, however. The number does not include benefits due to reduced medication retrieval time for nurses, increased utility, etc. because these are not actual economic savings. Other Notes on Validation A brief analysis of bed census data was performed to ensure that the period sampled was representative of the PICU and PCTU occupancy levels. Note that there are a maximum of 16 beds in the PICU and 15 beds in the PCTU. A summary of bed census data yielded: Table 8. Average occupancy slightly higher in 2008 study Sample Period PICU Average PCTU Average Total Average Jan. 1 Oct. 31, 2007 13.8 14.0 27.8 Oct 15. Nov. 1, 2007 12.7 12.9 25.6 Mar. 6 Mar. 26, 2008 14.7 13.4 28.1 Average occupancy was calculated by summing the number of beds occupied each day and dividing by the number of days in the sample. The baseline sample, January 1 to October 31, 2007, yielded an average of 27.8 beds occupied per day. This was 2.2 beds higher than during the 2007 Study (October 15 to November 27, 2007) and 0.3 beds lower than found by the BBCS team in 2008. Interestingly, the total average during all of 2006 and 2007 would have been 26.8 beds occupied per day. Because none of these figures differ by greater than 10%, the team determined that any changes caused by differences in occupancy would not affect results enough to account for the differences. Additionally, inclusion would complicate calculations and further understanding of results. Summary Each critical component of the study is outlined below: Nursing and Pharmacy Surveys The nursing and pharmacy surveys revealed important perceptions of the Medication Manager Pilot. In general the Medication Manager improved the communication of medication location and arrival for the nurses. This is shown through the significant increase in nurse knowledge of when medications arrive from the pharmacy as well as the location of these medications. 16

Nursing Medication Retrieval Time The Medication Manager increases the time nurses spend at patients bedsides by decreasing the frequency and duration of trips to the Medication Room and other retrieval channels. Quantitatively the Medication Manager reduced nurses trips to the medication room by 85% and and decreased the average time spent in the medication room by 42%. These percentages calculate to a four minute per bed per day overall reduction of non-value added nursing times. The monetary value of this savings however, is an unrealized $16,000. It is important to note that the $16,000 is not an actual savings but rather a transfer from non-value added work to potential effective work. Missing Medication Costs The findings of this report indicate that the Medication Manager Pilot is very helpful to the nursing staff. This is indicated by a reduction in medication retrieval time of 50% and a reduction in missing medication costs of about 85%. The reduction in missing medications may also impact the time spent making medications in the Pharmacy. However, it must be noted that the potential for savings from reducing missing medications, about $42,000, is much less than originally determined in the Fall 2007 Study. When compared to the Medication Manager Pilot costs, estimated to be about $100,000 per year, the economics are not as optimistic. Recommendations The BBCS team recommends that the Medication Manager System be standardized. Currently, Medication Managers operate independently with no standards. While they appear to be effective, the team believes that developing a standard work and identifying best practices will add to the improvements the Medication Managers effect in the PICU and PCTU. Suggestions include to: Improve medication request records: Currently Medication Manager record many medication requests on scratch paper when the requests are received. Organizing records by introducing a standard recording log could improve accuracy. Reduce channels for medication requests: Prior to implementation of the Medication Manager Pilot, most non-standard medication requests were sent to the Pharmacy through CareLink. Now, though, most requests are going through Medication Manager channels (via pager, phone, or in-person). The team recorded 342 CareLink requests in 14 days and 1,350 requests through the Medication Managers. 1,350 / (1,350 + 342) = 80% of requests went through the Medication Manager. These channels appear less conducive to accuracy than CareLink. Also, with multiple channels, there seems to be the opportunity for redundancy. Improve medication request/deliver trigger: Though missing medications have been reduced and nurses responded favorable (mean = 4.18) when asked how often they know when medications are delivered, there is still room for improvement. A signal to indicate a request or delivery may improve these metrics. The team has identified other areas for improvement. A greater emphasis on the Medication Administration Record (MAR) may help to enhance control over medications and reduce 17

wasted/missing medications. Also, continuing to keep records of missing medications (beyond this study) will help to track performance levels and improvements. Finally, once these suggestions have been implemented, we recommend reassessing the costs and benefits of the Medication Manager System. Considering only economics, the savings from reducing missing medications do no appear to cover the costs associated with the Medication Manager. These costs must be weighed against the benefits to reduced nursing medication retrieval time, nursing utility, and the potential for improved workflow. 18