Optimization of Facility Design and Workflow Processes at the Phlebotomy Clinic of Toronto General Hospital. Alexandra Barrett Tiffany Chow

Transcription

1 Optimization of Facility Design and Workflow Processes at the Phlebotomy Clinic of Toronto General Hospital Alexandra Barrett Tiffany Chow A thesis submitted in partial fulfillment of the requirements for the degree of BACHELOR OF APPLIED SCIENCE Supervisor: Professor Michael Carter Department of Mechanical and Industrial Engineering University of Toronto March 2008

2 EXECUTIVE SUMMARY This thesis concerns the Phlebotomy Clinic of Toronto General Hospital (TGH). Two areas of concern, identified by managers, were investigated: Scheduling and Workflow, and Facility Layout. In the Scheduling and Workflow section, a simulation was created to model the flow of patients and staff throughout the system. Data was obtained through measurement, interviews, and numerical inference. The results of the simulation identified two times of the day that have abnormally high patient wait times: Prior to10:30 and after 15:30, Monday through Thursday. In order to address these problem areas, three alternative scenarios were created and run through the simulation. In the first scenario, technicians were able to perform both phlebotomy and electrocardiograms (ECGs) on any patient, instead of being limited to the designation from their shift. Thus, instead of having two separate queues, patients were processed together. In the second scenario, the technician who usually handles preadmit patients assisted with outpatients when idle. For the third scenario, an additional worker was assigned to the clinic during the most problematic hours: 7:30 10:30, and 15:30 17:00. Additionally, each trial was run twice; once under mean historical arrival rates, and once under arrival rates generated as the high 90% of typical volume. While scenarios 1 and 2 showed moderate improvements of 4 to 5 minutes per average patient time in the system for the morning rush, they did not have much affect in the late afternoon. Scenario 3, on the other hand, showed strong improvements of over 7 minutes in the morning, and over 15 minutes in the afternoon. As a result, it was recommended that an additional resource be acquired during these key problem times. The facility redesign portion of this thesis addressed concerns related to space utilization. Problems like long patient wait times, crowding in high traffic corridors, lack of staff space, and staff/patient privacy were addressed. A modified Systematic Layout Planning (SLP) approach was used to assess the clinic s current space usage, operation levels, flow and activity patterns. From this analysis, a number of opportunities for improvement were identified. Two facility layouts were developed to change and improve current conditions. The improved layout offered solutions to the lack of staff space problem by consolidating two preadmission rooms i

3 into a new staff room. Staff lockers were moved into the new staff room to relieve congestion in the phlebotomy corridor. Congestion in the current ECG hallway from the wait area was moved into an ECG room located near the phlebotomy area. Due to the development of a new staff room, the temporary facility was converted back into an ECG room. The phlebotomy and lab dispatch areas were left intact. An aggressive layout involving several structural changes to the facility was offered to optimize work efficiency in the clinic. The ECG wait area from the improved layout was converted into a dual function ECG and phlebotomy room. The phlebotomy area was expanded into the laboratory dispatch area to accommodate an additional phlebotomy station and pneumatic tube transport system. As suggested in the improved layout, the old preadmit area and office were merged to form the new staff room, and lockers were subsequently moved there. To replace the old preadmit utility room, a new location was offered, adjacent to the staff change room. Both the improved and aggressive layouts were evaluated based on a qualitative criteria developed to assess the current as-is situation of the clinic. A preliminary evaluation of the proposed layouts with the clinic staff was found to be positive. ii

4 ACKNOWLEDGEMENTS The authors wish to acknowledge a number of key people who through their generous support made this thesis project possible. First, to Michael Caesar, a manager with the Shared Information Management Services (SIMS) Partnership at the University Health Network who provided the necessary contacts to initiate this project. Secondly, to Jay Hockley, Maria Amenta, and Maureen Marquette, senior administrators at the University Health Network who provided the opportunity to optimize their facility. Their resources, guidance, and frequent meetings with us were invaluable. We hope the results of our project will be of use to them. Thirdly, many thanks to the staff at the Toronto General Hospital Phlebotomy clinic, for sharing their valuable experiences with us and for allowing us to view their workspace. Finally, to our thesis supervisor, Professor Michael Carter, for his insight and support throughout the year. iii

5 TABLE OF CONTENTS 1 INTRODUCTION BACKGROUND MOTIVATION OBJECTIVES CURRENT STATE SITUATION PHLEBOTOMY AND ECG WORKFLOW SCHEDULING PRACTICES FACILITY LAYOUT DESCRIPTION PROBLEM DEFINITION SCHEDULING AND WORKFLOW PROBLEMS FACILITY LAYOUT PROBLEMS METHOD OF APPROACH PROBLEM IDENTIFICATION AND SCOPE DEFINITION DATA COLLECTION AS-IS MODEL OF WORKFLOW Scheduling and Workflow Simulation Development Patient Flow Patient Label Types Arrival Rates Task Duration Availability of Resources Scheduling and Workflow Simulation Validation AS-IS MODEL OF FACILITY LAYOUT Model Development Facility Model Validation TO-BE MODEL OF WORKFLOW TO-BE MODEL OF FACILITY LAYOUT SELECTION OF ALTERNATIVE FACILITY LAYOUTS AS-IS MODEL ANALYSIS SCHEDULING AND WORKFLOW SIMULATION FACILITY LAYOUT Current Space Usage Level of Operation Flow Analysis Activity Analysis Flow and Activity Relationship Space Requirements for Redesign SOLUTIONS AND RECOMMENDATIONS SCHEDULING AND WORKFLOW Scenario 1: All Staff May Perform Both Phlebotomy and ECG Scenario 2: Preadmit Technician Assists with Outpatients between Appointments Scenario 3: Additional Worker at Times of High Demand Comparative Analysis between Scenarios FACILITY LAYOUT RECOMMENDATIONS Improved Layout Description Evaluation of Improved Layout iv

6 6.3.1 Aggressive Layout Description Evaluation of Aggressive Layout CONCLUSION AREAS FOR FURTHER RESEARCH REFERENCES APPENDICES v

7 LIST OF FIGURES Figure 1: Floorplan of Clinic... 7 Figure 2: Crowded ECG Hallway Figure 3: ECG Room Converted to Temporary Staff Room Figure 4: Carts Occupying Dispatch Area Figure 5: System Flow Diagram Figure 6: Mean Patient Arrival Rates Figure 7: Heavy Patient Arrival Rates Figure 8: Observed Duration of Phlebotomy Procedures Figure 9: Sample Dummy Sequence to Account for Resource Breaks Figure 10: Comparison of Outpatient Arrival to Technician Availability Figure 11: Systematic Layout Planning Approach Figure 12: Modified Systematic Layout Planning Approach Figure 13: Average Time in System (Basecase and Mean Arrival Rates) Figure 14: Average Time in System (Basecase and Heavy Arrival Rates) Figure 15: Cluttered Dispatch Area Figure 16: Average Patient Time in System by Patient Type Figure 17: Procedure Volumes Figure 18: Average Wait Time for Phlebotomy and ECG Examination Rooms Figure 19: Comparing Average Time in System on Wednesday Figure 20: Comparing Average Time in System on Monday Figure 21: Comparing Average Time in System on Tuesdays Figure 22: Average Time in System, Scenario 1, Mean Arrival Rates Figure 23: Average Hourly Time in System Comparison, Basecase and Scenario 1, Mean Arrival Rates Figure 24: Average Hourly Time in System Comparison, Basecase and Scenario 1, Heavy Arrival Rates Figure 25: Average Time in System, Scenario 2, Mean Arrival Rates Figure 26: Average Hourly Time in System, Basecase and Scenario 2, Mean Arrival Rates Figure 27: Average Hourly Time in System Comparison, Basecase and Scenario 2, Heavy Arrival Rates Figure 28: Average Time in System, Scenario 3, Mean Arrival Rates Figure 29: Average Hourly Time in System Comparison, Basecase and Scenario 3, Mean Arrival Rate Figure 30: Average Hourly Time in System, Basecase and Scenario 3, Heavy Arrival Rate Figure 31: Average Hourly Time in System, Basecase and all Scenarios, Mean Arrival Rate Figure 32: Average Hourly Time in System, Basecase and all Scenarios, Heavy Arrival Rate Figure 33: Preadmit Office and Utility Storage Room Figure 34: Improved Layout Figure 35: Space for New Preadmit Utility Room Figure 36: Vacant Desk in Preadmit Area Figure 37: Blood draw Chair in ECG Room Figure 38: Outpatient Reception Area and Staff Change Room Figure 39: Aggressive Layout vi

8 LIST OF TABLES Table 1: Distribution of Patient Service Needs Table 2: Distribution Parameters Table 3: Shift Schedules Table 4: Areas for Redesign Table 5: Excel Model of Floorplan Table 6: Distances between Functional Units Table 7: Primary and Secondary Functions of each Clinic Area Table 8: Procedure Case Ratio Table 9: Estimated Weekly Flow Volume between Functional Units Table 10: Interrelationship Matrix Table 11: Reducing and Increasing Space between Functional Areas Table 12: Average Improvement of Patient Time in System Table 13: Room Specifications for the Improved Layout Table 14: Other Criteria for Evaluating the Improved Layout Table 15: Room Specifications for the Aggressive Layout Table 16: Other Criteria for Evaluating the Aggressive Layout vii

9 1 Introduction The Toronto General Hospital (TGH) Phlebotomy Clinic Thesis proposes solutions to a series of facility layout and workflow problems identified by staff at the outpatient facility. This project redesigns both the physical layout and process flows of the clinic using Industrial Engineering techniques. A combination of quantitative and qualitative approaches are used to analyze the current clinic setting, generate alternative processes and layouts, and evaluate their ability to address the identified problems. The overall aim of the redesign effort is to meet clinic needs, improve efficiency, and increase worker satisfaction. 1.1 Background Serving roughly five hundred patients every day, the Toronto General Hospital (TGH) Phlebotomy Clinic is a key component in patient care at TGH and the wider University Health Network (University Health Network, 2007). It is responsible for performing all phlebotomy and ECG tests on inpatients, preadmit patients, and outpatients who require them at the network. Inpatients refer to those who are assigned to the beds at the hospital. Since they are generally not well enough to visit the clinic, technicians perform sweeps of the wards twice a day. Although the same human resources may be used in these sweeps as in the clinic, they are not included in the examination of the clinic itself, since they fall outside its boundaries. Preadmit patients are those arriving on the day of a surgery or other intervention, and who require final tests before the procedure is carried out. Since numerous healthcare providers will meet with these patients, they are assigned to private rooms within the phlebotomy area and have technicians come to them for testing. Outpatients provide the bulk of admissions to the clinic, and as such will be the major focus for redesign efforts. These are patients who are not currently admitted to the hospital, but have physicians order for blood work or ECG testing. 1

10 Following are some key definitions related to this clinic: Phlebotomy: Phlebotomy, also known as venipuncture, refers to the removal of blood for blood tests. It is an extremely useful procedure for obtaining blood samples needed for many common medical tests. It is quick to perform, safe, and causes minimum discomfort to the patient. (Canadian Adult Congenital Heart Network, 2007) Electrocardiogram (ECG): A test used to monitor a patient s heart activity. The individual has a number of electrode pads placed on their body which measure electric current, and translate that information to determine the overall health of the heart. It is used to diagnose and monitor numerous heart conditions, and also to assess the health of a patient undergoing surgery. Like phlebotomy, it is safe, fast, and non-invasive. All technicians at the phlebotomy clinic may carry out both venipuncture and ECG procedures, although assigned duties vary day to day. (emedicinehealth, 2007) Pneumatic Tube System: A pneumatic tube system is a computerized medical transport system that sends specimens from various stations in the hospital to the core lab via capsules. The capsule that transports specimens is approximately six inches in diameter and can carry up to seven pounds. A station is to be installed in the outpatient clinic in the laboratory dispatch area by June (Swiss Logic, 2007) 1.2 Motivation The TGH phlebotomy clinic was identified over the summer of 2007 as an area where staff have expressed interest in having a formal optimization process take place. In particular, there have been concerns that the facility layout is not making full use of the space available, and is instead contributing to lower efficiency and worker dissatisfaction. Additionally, many feel that the human resources are not being used to their full advantage, as there exist predictable patterns of bottlenecks and idle time. (Outpatient Clinic Staff, 2007) 2

11 The need for attention is particularly timely at the moment, as one significant change to the clinic took place this year and another is scheduled for the upcoming summer. First, a portion of the outpatient section of the clinic was lost to an adjoining department. Most notably, a permanent staff meeting and lunch room no longer exists. Secondly, in order to reduce the time specimens wait to be sent to the lab for analysis, a pneumatic tube is to be installed. This will have the added effect of greatly reducing the space required for presorting specimens, since many of these tasks will now be carried out at the lab. As a result, part of the room currently used for collection will be free for other uses. 1.3 Objectives The main objectives of the Toronto General Hospital Phlebotomy Clinic thesis are twofold, reflecting the two different components of the thesis. First, the simulation and appointment scheduling half hopes to deal with improving process flow, staff scheduling and resource allocation through the clinic. At the same time, the facility layout component will attempt to create a solution to the current space problems by designing a layout that considers the space being lost and gained, the needs of the workers, and worker /patient satisfaction. The scheduling portion of the project will consider the following: i. Staff schedules {V1D, V2D, V3D, V4D, V5D, OPA, OPB, OPC, OPD, OPE, OP5, OP5E}, including break times ii. Measured patient arrival rates iii. Measured service time distributions iv. Patient flow through system The facility redesign portion of the project will consider the following rooms of operations in the outpatient area of the clinic: i. Laboratory Dispatch G-413 ii. Phlebotomy Stations G-417 iii. ECG Room G-418 iv. ECG Room G-419 3

12 v. ECG Room G-420 vi. Preadmit Office G-421 vii. Preadmit Utility Area G-422 viii. Outpatient Reception G-414 The laboratory managers and director of the Toronto General Hospital Phlebotomy Clinic have approved these following key performance indicators for evaluation of recommendations: i. Reduced wait times (hr) ii. Reduce bottlenecks (avg queue length) iii. Reduce traffic in crowded areas (# of people) iv. Meeting facility needs 4

13 2 Current State Situation The following are brief descriptions of three key components of the system: 2.1 Phlebotomy and ECG Workflow At present, an outpatient may arrive at any time during operating hours, and register at reception. If he/she does not have a Blue Card -used as identification within the UHN system- they fill out the necessary paperwork to obtain one before taking a seat in the waiting area. When a technician is available, they are generally assigned to the first patient in the queue, who is directed towards the venipuncture area with their order information. The exception is priority patients who have taken time sensitive medications and must be tested as soon as their wait time has elapsed. While most patients require blood work, some will also require an ECG test. If so, the technician will escort the patient after the completion of their blood work to the ECG area. Some patients only require an ECG and so flow directly there from the general wait room. For privacy, each ECG machine is in a private room. After all tests are complete, the patient leaves the clinic, while the technician records the specimen info, and deposits it in the collection area for further sorting and pickup. Once every 20 to 30 minutes, the specimens are moved to the core lab, where the actual analysis takes place (Operation Documents, 2007). Since the core lab receives samples from multiple sources in the hospital, it is considered outside the scope of this thesis, and will not be considered further. 2.2 Scheduling Practices This topic can be divided into two sections: the scheduling of patients, and the scheduling of employees. There are three separate schedule practices for the three different types of patients. First, inpatients are those in beds at the hospital who do not come to the clinic for their tests. Instead, technicians make rounds twice a day through the wards, performing venipuncture and ECG for those who require it. Since this occurs outside of the clinic, they will not be included in the scope of this thesis. Secondly, preadmit patients are those who arrive the day of a surgery and require final tests before interventions can be carried out. Their arrival times are scheduled 5

14 well in advance, and their progress through tests and visits from other medical technicians is continuously updated on a white board. Finally, outpatients form the bulk of arrivals to the clinic. They are unscheduled, have no appointments, and can arrive at any time throughout the day. Naturally, this leads to large bottlenecks at certain times of the day, while at other times clinic resources sit idle. Work shifts begin when the clinic opens in the morning. There is generally a single technician assigned to handle the preadmit patients, while the others work in the outpatient section of the clinic. The number of workers and their respective shifts is generally constant day to day, however the existence of predictable periods of either high patients wait times or worker idle time suggests that these distributions could be optimized. 2.3 Facility Layout Description The largest space in the phlebotomy clinic is the patient waiting area, a wide open room filled with chairs and located at the entrance. Along one side of this space is the reception area, where patients must sign in upon arrival. Behind this area, along the far wall, are entrances to private offices and the preadmit scheduling room. At one end of the wait lounge is a hallway leading to private offices and a number of medium sized meeting rooms, and a lunch room. At the other end of the waiting room, beside the reception desk, is the entrance to the venipuncture area. This room is tightly occupied and filled along three walls with blood collection booths. The fourth wall is the current location for staff lockers. Opposite from the exit to the wait lounge is a long, narrow hallway filled with doors to private ECG test rooms. The already narrow hallway is further cramped by the addition of chairs used by patients who have completed their blood work and are awaiting their ECG. A floorplan of the entire clinic is shown in Figure 1. 6

15 Figure 1: Floorplan of Clinic (Operation Documents, 2007) 7

16 Adjoining the blood collection area is a medium sized room where samples are placed and sorted after collection. A new policy, however, is to have many of these sorting activities carried out at the core lab. There are plans to install a pneumatic tube connecting to the core lab in the coming year, which will further eliminate the need for this room. Staff have expressed interest in having this space converted to a lunch/meeting room, since their existing space is being lost to an adjoining department next year. There are concerns, however, that some sorting processes, particularly those done for urine collections will still have to be carried out in the phlebotomy area since samples are not suited to pneumatic tube transport. As such, this space may not be suitable as a lunchroom. A number of general issues with the facility layout were identified after interviews with staff members. These problems generally fall under the issues of crowding in the venipuncture and ECG areas, the lack of space exclusive to staff, the lack of division between staff and patient areas resulting in privacy concerns, the lack of phlebotomy stations and ECG rooms during peak periods, and no dedicated storage space for porter carts. These facility related issues will be discussed in further detail in the problem definition section along with issues related to scheduling and resource allocation. 8

17 3 Problem Definition Consultation with staff and managers has led to the identification of key problems which this thesis hopes to resolve: 3.1 Scheduling and Workflow Problems i. Evaluate whether technicians should be assigned to either phlebotomy or ECG tasks for the day, or whether they should be free to take on whichever role is the most pressing. The current practice is to have all technicians work in phlebotomy except one who is responsible for performing most ECGs. A second technician, the backup, generally acts as a phlebotomist, but will also perform ECGs at times of high demand. However, observations have shown that many phlebotomists prefer to perform ECGs themselves when their patient requires one. Intuitively, allowing each technician to take on whichever role is most needed makes sense, but unless the improvements to wait times are substantial enough, it may not be worth the added difficulty of managing such an arrangement. ii. Consider whether the preadmit technician should assist in processing outpatients. Currently, a single technician is in charge of performing all phlebotomy and ECG tests to preadmit patients. On some days, few such patients are scheduled leading to long periods of worker idle time. During these periods, the technician could conceivably walk across the clinic to the outpatient area and assist in their duties. The priority must, however, remain on processing preadmit patients. iii. Determine whether there is a more optimal distribution for worker hours. According to worker accounts, the present employee schedule, when combined with random patient arrivals, lead to periods of idle time on Friday afternoon and substantially longer patient wait times in weekday AM time slots. For the most part, schedules are constant for each day of the week, and so could likely be better matched to when patients typically arrive. 9

18 3.2 Facility Layout Problems i. Crowding in the Venipuncture and ECG areas. There is a general sense of crowding in the venipuncture and ECG areas. Seven venipuncture stations are located along one corner of the clinic in crescent formation. A supplies station located in the centre of the crescent, provides easy access to supplies and the label printer but creates a tight walking space between stations. The high traffic hallway leading to the phlebotomy stations is also the location of staff lockers. Bottlenecks in that hallway develop when staff access the lockers and also when labels are being printed. The ECG rooms are located around the corner from the lockers. The hallway where the rooms preside is another source of crowding in the clinic. It is a high traffic area that staff and patients have access to. Patients requiring both phlebotomy and ECG must have the former done first. If there are no available ECG rooms after phlebotomy is completed, the patient must wait in the hallway leading to the ECG rooms. Crowding in this hallway is further compounded when there are wheelchair patients. Figure 2 shows the narrow hallway. Figure 2: Crowded ECG Hallway 10

19 ii. Lack of Staff Space. Staff working in the clinic have expressed concerns about the lack of private space. Currently, one of the three ECG rooms has been substituted as a temporary lunchroom, as the previous location of the staff lunchroom was lost to an adjoining department. The room is currently crowded with coaches and a refrigerator from the old lunch room. Figure 3 shows the current setup of the temporary staff lunchroom. Figure 3: ECG Room Converted to Temporary Staff Room iii. Location of Staff Space. Not only is a lack of staff space a problem, but so to, is the division of space. Currently, staff have to access their lockers in the busy blood collection area. As such, there are concerns over security as well as whether they are contributing to the crowding of the space. Another issue is that the substitute lunchroom and lockers are located far away from each other despite there being a high correlation of use between the two areas. Staff usually visit their lockers before and after breaks to change and retrieve/put back their belongings. Separating the two spaces over a long distance leads to congestion across the length of the crowded ECG hallway, and is also inconvenient for staff. Both the entrance of the lunchroom and staff lockers are located in very public places and in direct view to patients. Some staff feel that this lack of separation between their workspace and personal space makes it difficult to enjoy their breaks fully. An ideal location for both the lockers and 11

20 the staff lunch room would be one where the general public does not have access to, and yet is still easily accessible for employees. iv. Lack of Phlebotomy Stations and ECG Rooms. Phlebotomists performing inpatient rounds in the hospital come down to the clinic between 9:30am and 10:30am as added work resources to the outpatient clinic. However, these resources are not fully utilized, as at peak hours the number of technicians outstrip the number of available phlebotomy stations 11:7. Therefore, part of the As-Is Model analysis will determine what the ideal space requirement is for phlebotomy and integrate those results into the proposed layouts. ECG space requirements will also be investigated. v. Dedicated Storage Area for Porter Carts. Currently, six porter carts are being stored in the Laboratory Dispatch area temporarily, as there is no permanent location for storing these carts and the area is convenient for picking up samples. The carts currently add clutter and take up a fair portion of space in the dispatch area. Figure 4 shows how the carts take up space in the dispatch area: Figure 4: Carts Occupying Dispatch Area The pneumatic tube system, while, reducing the number of required porter carts, will not eliminate their function entirely, as urine samples will still be transported by cart. Therefore, the recommended layouts should consider a better way of storing these carts, whether 12

21 through another permanent location or through better organization of the current dispatch area. Integrating the space gained from the installation of the pneumatic tube system as well as better utilization of their existing space can provide relief to the facility problems identified. A thorough analysis of their current space usage will be completed in the As-Is Model Analysis section to determine opportunities for improvement. 13

22 4 Method of Approach This thesis project was divided into several stages: i. Problem Identification and Scope Definition ii. Data Collection iii. Modelling (As-Is and To-Be) iv. Evaluation of Alternatives 4.1 Problem Identification and Scope Definition Initial meetings were set up with key stakeholders to define the scope of the project. At these meetings, informal questions were asked pertaining to the current state of the laboratory and what was expected of the project. During tours of the phlebotomy clinic, the lab managers alluded to specific problem areas, like the crowded ECG hallway, the high volume of patients waiting in the reception area, the placement of the staff lockers, and the crowded setup of the venipuncture workstations. Scope definition was further refined after the completion of the progress report. Key performance measures for evaluation were agreed upon as well. 4.2 Data Collection Modes of data collection used in this project included reviewing operational documents, interviewing staff, reviewing historical data, and observing the clinical setting. Operational documents contained key pieces of information like preadmission clinic procedures, outpatient reception procedures, outpatient reception process flows, outpatient ECG services, hours of operation, and ward pickup times. Information such as the number of technicians working per hour per day of the week was obtained through staff interviews and schedule records. Historical data on outpatient arrival times and volume, preadmit arrival times and volume, and the number of patients requiring phlebotomy and ECG were manually inputted into Excel from patient records. Service times for phlebotomy and ECG were observed manually, as no such historical data was available. A facility floorplan of the clinic was provided by the Facility Management 14

23 Department but was not to scale. Manual measurements were taken to determine exact dimensions and also to verify space usage. 4.3 As-Is Model of Workflow Given the enormous amount of randomly distributed parameters to consider (patient arrival rates, patient service requirements, processing times, etc) simulation was the tool selected to represent the current workings of the clinic and test alternative scenarios. Simulations allow different scenarios to be tested visually on the computer without change to the actual operations of the clinic. It is particularly useful for gathering statistics on workflow processes and identifying and resolving bottleneck issues Scheduling and Workflow Simulation Development The first step in developing the simulation was determining the manner in which patients flow through the system. This was done by interviewing clinic staff, as well as observation to validate what had been communicated. This had to be done several times as different types of outpatients (those requiring phlebotomy, ECG, or both) follow different paths through the system. Once these paths had been determined, the duration of each stage was measured in one of two possible ways. For shorter tasks (typically <2 minutes), such as patient registration for a blue card, data relied primarily on estimates of the staff who routinely perform that function, but with a few measured observations to ensure the estimate was valid. While this is a less than ideal way to obtain numbers (taking many samples and fitting a distribution would be preferred), it many cases this was unavoidable due to time constraints and the scarcity of instances (only ~3% patients require a new Blue Card upon arrival). For the longer tasks (typically >5 minutes), namely the actual phlebotomy and ECG tests, numerous measurements were taken and fitted to a distribution to create a probability profile for service times. Once this profile for patients was created, staff schedules were compared to create a typical worker distribution for a single week. In addition, the break times assigned to each worker was included to create a realistic profile of the workforce at any point in a standard week. 15

24 The Simul8 platform used to create the simulation has a set of design elements which correspond to real world functionalities: 1. Work Items. These are the principle drivers of actions in a simulation, and are used to represent patients. These items are created, and passed around from one section of the simulation to the next, following the same path taken by a virtual patient. 2. Work Entry points. These are the locations where work items enter the system. The simulation uses the entry rate as a variable to reflect as closely as possible the actual arrival rate of patients. 3. Workstations. These are the physical locations where work items are processed, and correspond to every action carried out within the clinic, such as registration, phlebotomy, and ECG. 4. Queues. Since a workstation can only process a single work item at a time, queues are used when there is a line of patients waiting to be processed, such as in a wait room. 5. Work Exit points. These are located at the end of the system, and are used to remove work items once they have completed their flow through he clinic. 6. Resources. Resources are mobile units, much like a work item, which must be present at a workstation in order for it to function. These correspond to technicians at the clinic who are required to physically perform the phlebotomy and ECG tasks. (Concannon, 2004) Following is a more detailed description of the design of the key elements in the simulation 16

25 Patient Flow Figure 5 shows the path taken by patients through the clinic. Figure 5: System Flow Diagram Upon entering, patients join a queue for reception. Once they arrive at the front of the line, they may be directed to registration to acquire a Blue Card if they do not already have one, and then rejoin the queue for reception. Next, the patient enters the wait room. Although all patients wait in the same physical space, they may be there for one of three different reasons, each of which is handled separately in the simulation. First, under normal circumstances, a patient awaiting blood work (and the corresponding work item) sits in the wait room queue for the next available resource (phlebotomy technician) to take them to a phlebotomy station. In the second case, if the patient only requires an ECG, their work item is shown to be waiting in the ECG queue for the next available ECG technician and station. If a patient requires both blood work and an ECG, they will begin with phlebotomy and be sent to the ECG queue once it is complete. In a third 17

26 case, a patient may have taken time sensitive medication, and will require a blood test exactly 2 hours later. In these instances, the work item is sent to medication wait, where they are processed for anywhere from 0 to 2 hours (since many patients take the medication before arriving), before rejoining the general wait queue. In order to ensure expedited service, they are given a priority tag to receive service ahead of the patients without medication needs. After finishing with phlebotomy, or ECG, as needed, patients leave the clinic through one of two work exit points Patient Label Types Since patients enter the system with various service needs, a series of patient type labels were created to assign to each work item in the system. The three possible attributes identified were what tests they required (phlebotomy, ECG, or both), whether they required time sensitive blood tests, and whether they required registration for a Blue Card. Different methods were used for determining the ratio of patient types. Test needs were handled numerically, based on patient records: All patients entering the clinic were recorded in the main system, but those who also had ECGs were entered a second time in an alternate system. Subtracting these numbers yielded the number of patients who only had phlebotomy performed. Staff provided estimates on what ratio of ECG patients also require bloodwork, allowing for the distribution of patient service needs, by day of the week, shown in Table 1. Table 1: Distribution of Patient Service Needs identity Monday Tuesday Wednesday Thursday Friday blood, no medication wait 44.8% 42.7% 39.0% 41.9% 46.3% blood, medication wait 44.8% 42.7% 39.0% 41.9% 46.3% blood + ECG 3.1% 4.4% 6.6% 4.9% 2.2% ECG 7.3% 10.2% 15.4% 11.3% 5.2% total 100% 100% 100% 100% 100% 18

27 Note the sharp increase in proportion of ECGs on Wednesday. After consultation, it was discovered that there is a cardiology clinic run on this day elsewhere in the hospital, leading many patients to be directed here for tests. In the case of time sensitive blood tests, numbers were obtained from estimates by technicians and receptionists. They averaged that half of all strict phlebotomy (no ECG) patients require these medication imposed waits, followed by queue priority (shown in Table 1). The duration of the wait can last anywhere from 0 to 2 hours, depending on how long before arriving the patient took their medication. According to estimates, the distribution is completely random, since some people do not take their medication until arriving at the hospital, while others plan ahead and time their arrival exactly 2 hours after administering their dose. Both these instances were observed during a short validation stage. As a result, the length of medication wait was set to a Uniform probability distribution from 0 to 120 minutes. Finally, through interviews with the receptionist, it was estimated that only 1 in 30 patients arriving at the clinic do not have a Blue Card and require registration. Since this is a fairly rare occurrence, validation through sample observations did not take place Arrival Rates In order to determine patient arrival rates, records hand written by the receptionist for any time within the prior year (2007) were available for mining. However, since some times of the year experience lower volumes than usual, staff were asked what months they felt would provide the most reasonable representation of a typical, heavy volume of patients. The final decision was to compile data from the months of September, October, and November of 2007, as they fell between the summer and winter holidays. Earlier spring months, while viable, were not chosen due to the time constraints of manually entering each arrival time. The average number of patients arriving in each hour of each day of the week was compiled, and fitted with an exponential distribution. While a better practice would have been to use software such as BestFit to determine the closest matched distribution, the literature convention for random arrival times (exponential) was used since the times recorded for each patient were when the receptionist 19

28 entered them into the system, not when they first arrived in the clinic. (Concannor, 2004). As a result, if a pair of people entered at once, the record would show them as arriving at least a minute apart to account for the time it took to process the first. Once compiled, the data shows the variation in number of patients from one hour and day to the next. Figure 6 shows these incoming numbers. Average Number of Patients Arriving per Hour, Mean Arrival Rate Number of Patients (per hour) Monday Tuesday Wednesday Thursday Friday Hour of the Day Figure 6: Mean Patient Arrival Rates Note that the highest arrival rates are before noon, and tend to peak around 9am-10am. Friday has by far the smallest volume of patients, while Mondays are the most busy in the AM period. While these average arrival rates are certainly useful, an alternative heavy profile was also created to see how the system would fare during a particularly busy week. A heavy profile was arbitrarily chosen to be one with a greater volume of patients than 90% of similar dates. This was 20

29 accomplished by first taking the average number of patients seen on each day of the week for September to November, 2007, and calculating the corresponding standard deviation (σ) of this value. Next, by assuming a Normal probability distribution for the number of patients to arrive on a given day, each σ was multiplied by 1.28 and summed with the average, according to P(0.9) = μ + (1.28)σ, for Normal distributions Calculations are shown in full in Appendix A. This new value represents a volume of patients in the top 90 th percentile of historical data, assuming a Normal distribution. These volumes are represented in Figure 7 per hour of day and day of week. Average Number of Patients Arriving per Hour, Heavy Arrival Rate Number of Patients (per hour) Monday Tuesday Wednesday Thursday Friday Hour of the Day Figure 7: Heavy Patient Arrival Rates 21

30 The distribution controlling the arrival rate of work items was set up to sample average interarrival time (λ) according to the time in the simulation, and automatically adjust each hour. The exact lambda values used to create exponential arrival rates are provided in Appendix A Task Duration Task durations were either determined through measurement or by asking staff for estimates. While the former is vastly preferred from a statistical standpoint, it was either impossible to observe enough instances of each task to create a subjective measure, or else short enough of a task to make little to no difference in the overall model. The largest data gathering initiative was done for timing phlebotomy procedures. Figure 8 shows the observations taken. Frequency of phlebotomy duration Observations Duration (minutes) Figure 8: Observed Duration of Phlebotomy Procedures 22

31 Using BestFit software, the histogram above was found to most closely match a Pearson VI distribution, which was used in the simulation to generate random times for all phlebotomy procedures. Measuring the typical length of an ECG procedure was more difficult, since fewer occur during the day. Instead, only 5 times were collected (Appendix A), with a mean time of 8.6 minutes, and a standard deviation of just.55 minutes. Staff confirmed that ECG procedures have much less variance in their duration than with phlebotomy, since all take a fixed amount of time at the machine, meaning the only variable is how long it takes for the patient to partially undress and have the sensors attached. Additionally, when asked for an estimate of how long they tended to take, staff either answered 10 minutes, or in one case, 5-10 minutes. The distribution was assigned to normal to account for the symmetry (lack of outliers), given a mean time of 9 minutes (slightly higher than observed to account for the general consensus of 10 minutes among staff), and a rounded up, but still small, standard deviation of 1 minute. The short duration of reception tasks made timing with reasonable precision difficult. Instead, a Normal distribution centered around 1.5 minutes was chosen, according to interviews with staff. The standard deviation was assigned to 25% of the mean, or.375 minutes. Since every patient will go through this process, any variation to this (already small) time span would not have an appreciable effect. For registration, only a single observation was made due to the rarity of events. In that instance, the duration was 2 minutes. The worker commented that this was normal for an English speaking patient, but that if they were communicating through an interpreter, or had poor hearing, it could take a bit longer. The estimate of an average time of 3 minutes was taken, and fitted to a normal distribution with standard deviation of 25%, or.75 minutes. Table 2 summarizes data used: 23

32 Table 2: Distribution Parameters Task Phlebotomy ECG Reception Registration Duration (minutes) Pearson VI distribution α 1 : α 2 : 4.44 β: 1.83 Normal distribution μ: 9 σ: 1 Normal distribution μ: 1.5 σ: Normal distribution μ: 3 σ: Availability of Resources The 12 possible shifts to assign to a technician, including all break times and designation, are as follows: Table 3: Shift Schedules Shift Designation Start Break 1 (15 minutes) Lunch (30 minutes) Break 2 (15 minutes) End V1D Phlebotomy 9:30 11:30 13:30 13:45 V2D Phlebotomy 10:30 11:30 13:30 14:30 V3D Phlebotomy 10:30 11:30 13:30 14:30 V4D Phlebotomy 10:30 11:30 13:30 14:30 V5D Phlebotomy 10:30 11:30 13:30 13:45 OPA Phlebotomy 7:00 9:00 12:30 13:45 15:00 OPB Phlebotomy 7:30 9:20 12:30 13:30 15:30 OPC Phlebotomy 12:30 14:00 15:30 OPD Phlebotomy/ ECG (backup) 8:30 8:45 11:30 13:45 15:00 OPE ECG 7:30 9:45 12:30 14:15 15:30 24

33 OP5 OP5E Phlebotomy* Phlebotomy* 9:00 11:00 13:00 15:00 17:00 9:30 11:00 13:00 15:00 17:30 * When OPE, the usual ECG technician, ends at 15:30, OP5 and OP5E perform both tasks as needed for the rest of the day While shift patterns vary from one week to the next, the general norm is to have one resource on each shift working every day, with the exception of SP5 and SPD5: For Monday through Thursday, two staff work as SP5, and none as SPD5, while on Friday, they both work as SPD5 to accommodate the earlier closing of the clinic. There is also one additional worker assigned as OPSN, or the senior technician. However, since this staff member tends to float between the lab and the clinic, they cannot be considered as a full time resource. Instead, they have been discounted from the simulation, and may be considered as a possible resource for remediation of any problems identified. While the software allowed for detailed scheduling to account for the beginning and end of each shift, break times had to be handled separately. This was done by creating an individual dummy sequence for each resource, consisting of a work entry point, leading to a queue, leading to a workstation, leading to a work exit point. Figure 9 shows this sequence for the OPA break. Figure 9: Sample Dummy Sequence to Account for Resource Breaks Each entry point had scheduled arrivals corresponding to when a break or lunch is to take place for that shift. Upon entering the queue, a Visual Logic (VL) event is called to ensure that the resource will prioritize the new break event over any work items representing patients. In order to run through the workstation, the appropriate resource needs to be present for a duration of time equal to either 15 or 30 minutes, depending on the nature of the break (also handled through VL). This technique ensures that a) a resource will not quit a patient in the middle of 25

34 a procedure to take a break, and b) if the resource has to start their break late, they will push back their return to ensure they receive their full allotted time. As a final consideration, in order to properly represent human resources in a simulation, there should be some recourse to account for the fact that humans do not work nonstop like a computer, but rather take short breather breaks throughout the day. Observations showed plenty of reasonable actions such as drinking water, washing hands, speaking with a co worker, or even walking between stations (since Simul8 cannot handle travel times of resources, only work items). This was represented in the simulation by assigning an availability percentage to all workers of 85%, meaning out of 10 minutes, each worker is typically inactive for 1.5 minutes Scheduling and Workflow Simulation Validation Two forms of validation had to take place. First, as already alluded to above, the input data had to be screened to ensure it represented the real world situation. In cases where data was collected from interviews, a few measurements were collected to ensure the estimate was reasonable. Although these data points represented the least certainty (having not been formally measured and assigned a distribution), they were also the most minor due to their short duration and/or scarcity of use. For the more significant, measured, distributions, validation took place informally by having staff estimate the typical duration of these tasks. Additionally, a copy of a report done on the Mount Sinai phlebotomy clinic (MacDonald, 2007) showed a similar average time to what had been observed at TGH. The second stage of validation took place once the simulation of the clinic was complete. The data used to validate the model was a patient s average time in the system (excluding medication imposed waits), as compiled for each hour of each day of the week across ten weeks (over 10,000 data points in total). Since it is unfeasible to measure a patients time in the system for every hour on a given week, and even more so to track enough to create a typical profile, two alternate methods were used to ensure the simulation was properly modeling the real world situation. First, interviews with clinic staff were conducted to identify the periods they perceived 26

35 as having the longest patient wait time. All staff had a consistent view of what times of the day tended to form the longest patient queues, and could often cite times when they had to call in additional help to deal with the high demand. The second method used was to examine patient s arrival rates, and compare it to the number of technicians working across each hour of the day and each day of the week (Figure 10). Since the volume of patients on Friday was significantly lower than for every other day of the week (Appendix B), and staff had confirmed the decrease in demands on that day, their data points were omitted as outliers. Comparison of Demand to Resource Allocation, Monday to Thursday average workforce (staff) Staff Patients average # of arrivals (patients) Hour 0 Figure 10: Comparison of Outpatient Arrival to Technician Availability This method identified two key locations where the volume of patients was disproportionately higher than the number of workers; prior to 10:30, when the majority of technicians return from inpatient sweeps, and after 15:30, when all but two technicians finish. These were the same times that staff had known there to be larger queues, and were the same times identified by the simulation results as having the longest patient wait times (see Section 5.1 for full analysis). 27

36 While this methodology validates the simulations ability to locate periods of long wait times, it is impossible to determine whether it accurately predicts the duration of the waits; that is to say, there may be an issue with scaling. Under ideal circumstances, a lengthy collection period of tracking patient time in the system could be used for validation, but under existing time constraints, the existing results must suffice. 4.4 As-Is Model of Facility Layout A facility layout has four fundamental elements, functional units, unit relationships, space, and constraints and is defined as an arrangement of functional units in a given space with the objective of accommodating unit relationships under a set of constraints (Garcia-Diaz, 2007). A modified version of the Systematic Layout Planning (SLP) approach was used for analyzing the current layout of the clinic Model Development SLP is a widely used approach to facility planning originally used for designing factory layouts. It is a systematic method of generating feasible alternative layouts by considering the flow of resources and resource activity between functional units in the facility. (Garcia-Diaz, 2007) Resource flow analysis involves analyzing staff traffic between functional units and resource activity analysis involves qualitative ratings that consider both flow preference and constraints. Figure 11 outlines the SLP approach visually: 28

37 Input Data and Activities 1. Flow of Materials 2. Activity Relationships 3. Relationship Diagram 4. Space Requirements 5. Space Available 6. Space Relationship Diagram 7. Modifying Consideration 8. Practical Limitations 9. Develop Layout Alternatives Figure 11: Systematic Layout Planning Approach (Garcia-Diaz, 2007) The SLP approach consolidates the resource flow analysis with the activity analysis in a space relationship diagram. Functional units to scale are connected together with lines that vary in thickness according to the closeness rating given to pairs of units found in the interrelationship matrix. Space requirements and availability are labeled for every functional unit. Alternative layouts are then generated from the space relationship diagram by rearranging the functional units within a scaled drawing of the facility boundaries. Additional considerations and practical limitations are integrated in the process. 29

38 In comparison, Figure 12 shows the modified SLP Approach for the redesign of the TGH clinic: Input Data and Activities 1. Space Usage 2. Levels of Operation 3. Flow of Resources 4. Activity Relationships 5. Space Requirements 6.Develop Layout Alternatives Figure 12: Modified Systematic Layout Planning Approach The modified SLP differs from the conventional approach because space requirements and practical limitations are determined before conducting the flow and activity analysis. This way, workflow procedures are known before conducting the flow analysis. Also, resource flow estimates can take into account operational volumes of the clinic. Step one, Space Usage was assessed by identifying functional units in the clinic, determining the physical dimensions and distances between units, and classifying the primary and secondary functions are of each unit. How the space was primarily being used was considered the primary function of the unit and additional uses of the space were considered secondary functions of the unit. By classifying primary and secondary functions of each unit, opportunities for reallocation and movement were found. 30

39 Step two, Levels of Operation, involved analyzing current operation volumes. Information like the daily average volume of patients for each procedure type, the number of personnel staffed each day of the week and their schedule, and estimated wait times for phlebotomy stations and ECG examination rooms, was useful for establishing space requirements. The modified SLP approach also eliminates the need for a space relationship diagram. Instead, travel distances defined in the current space usage step were compared with their respected closeness rating in the interrelationship matrix to determine if units were spaced ideally. Any violations were then reviewed in more detail and then considered for change in the alternative layout. Conversely, efforts were made to preserve functional pair distances that matched their closeness rating. For example, if it was desired that two functional units should be far apart and the two units were currently far apart, then efforts were made to preserve their location. The last step, determining space requirement for each functional unit was estimated from information gathered in the Level of Operation and Flow Analysis steps. As mentioned before, operation volumes provided evidence for increasing and reducing the size of functional units. Analyzing congestion in certain units or corridors, as revealed by the flow analysis, was also useful for determining space requirement. The culmination of steps in the modified approach yielded insight into how space should be better utilized in the clinic, something that was lacking in the conventional approach Facility Model Validation Model validation involved measuring whether or not the model accurately represents the physical system of study (Francis, 1992). One way that validity was measured was comparing the results of the model with actual system performance. While this was straightforward to do with simulation models, facility models were harder to validate. The first step to validation involved reviewing assumptions of the model. Making assumptions involved trade-offs between the degree of realism of the model, the ease of manipulation, and clarity of results (Francis, 1992). A rule of thumb used was to start with a simple model and further refine until the required realism and results was achieved. For the purposes of this project, a simple model consisting of a 31

40 facility layout, travel distances and flow between functional areas, and an interrelationship matrix were developed to analyze the relative merits of the layouts based on a qualitative criteria. The model will be validated with lab managers to see if results accurately portray the working patterns of the clinic and subsequently, the ratings given to each pair-wise functional relationship. 4.5 To-Be Model of Workflow Once the basecase (or as-is) model had been generated and areas of difficulty identified, the next step was to create and test alternative scenarios to improve on the key performance measures. Three alternative scenarios were compiled based on suggestions from management and system observation, and directly address the problems outlined in section 3.1: i. Allow staff to perform both phlebotomy and ECG tasks throughout the day At present, only one technician is assigned to performing ECGs, with a second backup technician who performs both ECG and phlebotomy duties. All others are strict phlebotomists. However, some technicians already prefer to perform ECG on their bloodwork patient, rather than changing the individual over to a new staff member. This trial is to test whether allowing each technician to perform both tasks will have any appreciable improvement on patient time in system. This scenario required minimal modifications to the basecase. The only difference is that the constraint restricting which resources function at the phlebotomy and ECG workstations has been removed, allowing any technician to process a work item anywhere. ii. Have the Preadmit technician assist with outpatients between scheduled appointments There is at present a single technician in charge of handling both the phlebotomy and ECG work for all preadmit patients. The number of such patients ranges from 10 to 25 a day. The time between patients is currently idle, leading to management s suggestion of having the technician assist with outpatient when they have the time. 32

41 In order to modify the basecase, an additional resource entitled Preadmit was added to the simulation. This worker was assigned the ability to perform both phlebotomy and ECG procedures within the outpatient area. In order to simulate the technicians preadmit duties, a dummy system, similar to those used for controlling breaks, was set up outside of the usual clinic system. At prescheduled times, a work item would enter this dummy system and require the technician to attend to it with a priority over any outpatient. Only once all such patients were cleared from the queue would a Visual Logic (VL) event be called to let the resource access the outpatients again. In order to determine the number and frequency of preadmit patients, and the duration of their tests, records for the weeks of February 11th and February 18 th, 2008 were recorded. Since the arrival of patients it s scheduled in advance, rather than bring random, and only two weeks of data were collected, it did not seem reasonable to fit arrival rates to a distribution as was done with outpatients. Instead, a schedule of arrival times was coded into the system, such that the pattern would be the same for each weekly trial of the simulation. Both weeks for which data was available were extremely similar in terms of overall patient volume on each day, and the times of arrival. In order to keep this scenario as conservative as possible, for each day of the week the heaviest of the two available options was taken. The final schedule is shown in Appendix A. To create the distribution for the duration of the phlebotomy and ECG tests, the data collected was fit to a normal curve. Bestfit was not a good option since the majority of data points had been rounded to the nearest 5 minutes. It was also unreasonable to simply sum the existing distributions for outpatient tests, since the circumstances are quite different: unlike outpatients, preadmit patients have both tests performed at once, and have already had several levels of pre registration take place to streamline the process. As a final consideration, the availability % of this worker was lowered from 85% to 75%. This was done for two reasons: First, to represent time lost when traveling between the preadmit and outpatient sides of the clinic, and secondly, because while the simulation automatically directs the resource to where it is needed, in the real world the technician would not know whether it were needed on the other side of the department without walking back to check. 33

42 iii. Assign additional workers to times with the longest patient queues As previously discussed in section 4.3.2, at present there is a strong disparity between the number of patients arriving at the clinic and the number is resources available for the hours before 10:30am and after 3:30pm. As a result, these hours consistently have the longest patient time in system (see Section 5.1). In this scenario, the effect of having even a single extra worker available during these times is observed to see how large of a potential impact it could make. The basecase was modified by adding a new resource who could perform both phlebotomy and ECG tasks, and having them work between 7:30am and 10:30am each morning, and 3:30 to 5:00 pm each afternoon. Because of the short duration of shifts, no breaks were included. 4.6 To-Be Model of Facility Layout Two alternative layouts improved and aggressive were generated using an iterative approach that started with the current layout. Modifications to the current layout were done subjectively based on visual judgment and best practice guidelines from hospital architecture books. Suitability of alternatives was determined through qualitative visual judgment and was guided by mathematical analysis. Travel distances were recalculated and then compared with the interrelationship relationship matrix. 4.7 Selection of Alternative Facility Layouts The laboratory managers of TGH requested for two alternative layouts to be generated: one depicting an improved layout that considers the physical constraints of the facility (including the proposed location of the pneumatic tube system) and second, a more aggressive redesigned layout developed under relaxed constraints. The final layout which is likely to be a consolidation of both the improved and optimal layouts will ultimately be decided by the key stakeholders. Both the descriptive quantitative model and qualitative interrelationship model will be used to evaluate the end layout configuration. Other qualitative criteria like ease of implementation, flexibility of layout, space utilization, employee satisfaction, may also be considered, depending on the preferences of the stakeholders. 34

43 5 As-Is Model Analysis The Toronto General Hospital Phlebotomy Clinic is currently experiencing inefficiencies associated with its use of resources and physical space. Long patient waits during peak hours in the morning, lack of workstations to the handle high patient volume, and loss of physical space to an adjoining department were some of the problems identified previously in this report. This section models the current state of the facility using a simulation model and a modified systematic layout planning model to yield insights into the root causes of these problems and provide a basis for recommendations in the to-be section of this report. 5.1 Scheduling and Workflow Simulation The simulation previously described in section 4.3 was run twice over a 10 week period, with each patient s time in system (excluding medication imposed waits) recorded according to the hour and day they first entered the clinic. This includes both their wait times, and service times. The first run was performed using average patient arrival rates. The values collected are given in Appendix B, and shown in Figure 13, below: 35

44 Average Time in System - Basecase, Mean Arrival Rate Duration (minutes) Monday Tuesday Wednesday Thursday Friday Hour of the day Figure 13: Average Time in System (Basecase and Mean Arrival Rates) Note that the times are highest in the morning, when the arrival rate peaks, but the number of workers has not reached full capacity (see Figure 10). Additionally, there is a second jump in wait times at the end of the day, when the majority of staff have left for the day (Figure 10). Fridays do not appear to have any serious issues with wait times, as the patient arrival rates are much lower (Figure 6). When the heavy patient arrival rate was used (top 90%), the resulting duration of stay in the system increased, as given in Appendix C, and Figure 14, below 36

45 Average Time in System - Basecase, Heavy Arrival Rate Duration (minutes) Monday Tuesday Wednesday Thursday Friday Hour of the Day Figure 14: Average Time in System (Basecase and Heavy Arrival Rates) The general patterns of patient duration in the system are repeated for the heavy arrival rate, although they are magnified. The only obvious exception to the rule is that late afternoon Friday visits begin to have increasing waits. Since the key areas of interest across both arrival rates are prior to 10:30 am, and after 3:30 pm, Monday to Thursday, the three alternate layouts presented in Section 6 will be evaluated based on their ability to improve upon these specific times. 5.2 Facility Layout Using the modified SLP approach, the current conditions of the facility layout were analyzed. Current space usage was analyzed by identifying the primary and secondary functions associated with each unit, the amount of space currently occupied by each unit, and the distances between units. Operation levels of the facility were determined based on data generated by the simulation. Staff flow rates between functional units were estimated based on observations 37

46 conducted at the clinic. A qualitative, interrelationship diagram was created showing the closeness ratings between functional unit pairs. Closeness ratings were scaled from 1 to 5 with low numbers indicating a preference for separation and high numbers indicating a desire for closeness. In the next step, the distances between functional units were compared with the interrelationship closeness ratings to determine areas for improvement. Further opportunities of improvement were identified when the primary and secondary functions list was revisited. Secondary functions that did not logically belonged to the assigned unit were considered for relocation or grouping into a new functional area. This led to the final step of determining space requirement for each functional unit. Depending on results from the flow and operation analysis, space usage for each unit was left the same, reduced or increased Current Space Usage A floorplan of the outpatient clinic was used to provide geographic orientation of the facility. Table 4 shows the outpatient areas G-413, G-115, and G-418 to G-422 for redesign: Table 4: Areas for Redesign Area Laboratory Dispatch G-413 Phlebotomy Stations G-417 ECG Room G-418 ECG Room G-419 ECG Room G-420 Preadmit Office G-421 Preadmit Utility Area G-422 Outpatient Reception G-414 Dimensions (sqt) length 27ft height 17ft length 27ft height 17ft 10x11 10x11 9x11 8x10 8x10 13x15 38

47 Room G-410 was confirmed to be out of scope for redesign since it belonged to the preadmit staff. The preadmit utility area G-422 was available for redesign on the condition that preadmit services could be integrated into the space. Office G-421 is currently being used by a preadmit staff worker but is considered within scope so long as the functions of the space is preserved elsewhere. After dimensions of the space were assessed, a scaled model of the floorplan was developed in Excel to determine the rectilinear distances between functional units in the clinic. ECG rooms G-418 and G-419 were consolidated into one functional area. ECG room G-420 was the location of the temporary staff room. Staff lockers at the time of observation were located in the hallway leading to the phlebotomy area across from the service station. Preadmit areas G-421 and G-422, while within scope for redesign, were excluded from distance calculations since these areas were not being accessed by outpatient staff at the time. The location of these two rooms were highlighted in grey. Bathrooms and hallways were highlighted in orange and purple respectively to signify their location. Table 5 shows a screenshot of this model: Table 5: Excel Model of Floorplan Legend: A: Laboratory Dispatch B: Phlebotomy Stations C: Reception D: ECG Rooms E: Staff Room F: Lockers 39

48 Since the largest areas of the clinic, laboratory dispatch and phlebotomy, were accessed heavily around their weighted centriod, it was decided that the distance between functional units should be taken from the centroid rather than from the perimeter. For example, in the phlebotomy area, the service station is located approximately at the centroid. Each time a patient was served, the phlebotomist would wash their hands, pick up supplies, and print labels from the service station. Since the phlebotomy stations were arranged around the perimeter of the room, the average path traveled would work out to be roughly around the centre of the area. Another example supporting the use of the centroid is the laboratory dispatch area. The centroid of the laboratory dispatch area is located near the storage counter where blood samples are dropped off from the phlebotomy stations. This further supports the use of the centroid as a means for estimating the average destination point of the area. Table 6 shows the calculated distances between functional units in the clinic. Each cell of the matrix is the distance in feet between functional areas corresponding to the cell computed. Table 6: Distances between Functional Units Rectilinear Distance Functional Area A B C D E F A: Laboratory Dispatch B: Phlebotomy Stations C: Reception D: ECG Rooms E: Staff Room 26.5 F: Lockers After several visits to the clinic, the primary and secondary functions were identified for each room in the outpatient area. In addition to problems cited before about the lunchroom from the staff, it was found that the laboratory dispatch area and the preadmit utility room could be better organized. A summary of these findings can be found in Table 7: 40

49 Table 7: Primary and Secondary Functions of each Clinic Area Area Primary Function(s) Secondary Function(s) Laboratory Dispatch G-413 Phlebotomy Stations G-417 ECG Room G-418 ECG Room G-419 ECG Room G-420 Preadmit Office G-421 Preadmit Utility Area G-422 Outpatient Reception G-414 specimen pick-up and storage area performing venipuncture performing ECG tests performing ECG tests staff lunch room permanent office for preadmit worker storing supplies for preadmit workers and reception desk registering outpatients cart storage supplies storage specimen drop-off window pneumatic tube location furniture storage (TV, unused ECG bed) venipuncture supplies storage label printing sinks lockers calling outpatients from waiting area The primary function of the laboratory dispatch area was for storing porter carts and dropping off specimens from the phlebotomy area. However, the space was also being used to store abandoned furniture and equipment. Figure 15 show how the space in the dispatch area is being used currently: Figure 15: Cluttered Dispatch Area 41

50 The L-shaped room is cramped, with counters surrounding the perimeter. Upwards of six carts are stored in the room daily, taking up valuable walking space in the clinic. There is no alternative location for these carts in the clinic. The preadmit utility storage room also needs to be reorganized and cleaned out. Old office equipment was found on the counters, and cabinets were not being used to their full capacity. Consolidating items in these two areas could free up valuable space. One positive aspect of the current setup was the phlebotomy stations. While space is limited in the area, the crescent formation of workstations around the service counter allowed for efficient movement. The close proximity of the phlebotomy area to the ECG room was also desirable. Their relative locations facilitated a smooth workflow for staff attending patients requiring both bloodwork and ECG testing. Efforts were made to preserve their layout and proximity Level of Operation Clinic operations were analyzed to identify inefficiencies attributed to the spatial layout. To determine space requirements and constraints, information on current patient load and staffing patterns were analyzed. By running the as-is simulation model over a period of ten weeks, statistics were gathered on the average procedural volume per day, patient time in the system, wait times for phlebotomy stations and ECG examination rooms, and the number of staff on duty. Figure 6 shows that the clinic serves roughly 200 outpatient per day. Monday and Tuesday are the busiest days of the week. Workload volumes taper off towards Friday. During peak times, roughly thirty patients arrive at the clinic per hour. The facility can service up to seven phlebotomy patients and two ECG patients at a time. However, looking at the staff schedule it appears the number of resources on hand rather than space availability is the limiting factor to efficiency (see Figure 10). The number of total staff available to work through the peak volume is less than the number of workstations available in the clinic. During peak hours of 8:00 to 10:30am, only three to seven staff are on duty to see patients. The lack of staff to meet the higher patient demand results in 42

51 more time spent waiting by the patient. To support this claim, Figure 16 shows the average patient time in the clinic from registration to discharge, as separated by patient type. Data values are provided in Appendix D. Average Time in System Minutes Blood w/ Med Blood w/o Med ECG Only Blood + ECG Average Hour of Day Figure 16: Average Patient Time in System by Patient Type Of the staff available to meet the 8:30-10am peak period, one is designated to handle ECG cases, another acts as a floater who can handle either phlebotomy cases or ECG, and the rest deal with phlebotomy cases only. To determine if the staff allocation and ratio of phlebotomy stations to ECG examination rooms are appropriate, the ratio of patient types were calculated from historical data. Table 8 summarizes these calculations. 43

52 Table 8: Procedure Case Ratio blood blood+ecg ECG only Monday Tuesday Wednesday Thursday Friday These procedure case ratios were used to extrapolate the procedure type volumes provided in Appendix D, and shown below: Procedural Volumes # of Cases Blood Only ECG Only Blood + ECG Total Monday Tuesday Wednesday Thursday Friday Day of Week Figure 17: Procedure Volumes The majority of patients require phlebotomy only. On Wednesdays, higher volumes of ECG cases are performed since a cardiac clinic runs elsewhere in the hospital. With Wednesday data excluded, the average ratio of patients that come in for phlebotomy only, versus ECG, is roughly 9:1. Since the average service time for phlebotomy (just over 7 minutes) is slightly faster than the average service time for ECG (around 9 minutes) the 7:2 ratio of phlebotomy stations to ECG stations appears to be reasonable. 44

53 However, given the inconsistent service times for phlebotomy (depending on the number of vials of blood required) and inconsistent type of patient mix throughout the day, there should naturally be some variability in the wait times per procedure per hour. Figure 18 shows this variability of wait times throughout the day excluding medication wait and procedure time, according to 10 runs on the simulation model, using mean arrival rates. Full simulation output is shown in Appendix D. Average Wait Time Wait Time (min) Blood w/ Med Blood w/o Med ECG Average Wait Time Hour of Day Figure 18: Average Wait Time for Phlebotomy and ECG Examination Rooms It appears that there is higher wait time for phlebotomy in the morning and for ECG mid day. This could be attributed to two interrelated factors: the number and allocation of staff per hour and the number of available phlebotomy stations and ECG rooms. Since the number of phlebotomy staff available from 8-10am does not exceed the number of available phlebotomy stations, the high wait time in the morning for phlebotomy is attributed to staff volume. Mid day, the wait times for phlebotomy are at its lowest while wait times for ECG are at its highest. This means that the ratio of available ECG staff and phlebotomy should be altered to cater to this 45

54 change in demand. So while the 7:2 ratio of phlebotomy stations to ECG rooms may be appropriate on average, the ratio should change according to demand throughout the day. As a result, further investigation into a floater room may be worthwhile Flow Analysis The flow pattern in the clinic is circular. The reception area provides the two entrance points into the clinic area. Staff can circulate clockwise from reception to laboratory storage to phlebotomy, or counterclockwise from reception to the staff lunchroom and ECG area. Since staff accompany patients into the clinic, travel between functional units at the clinic is for the most part, a direct result of workflow sequence. Patients are called from the reception desk and escorted to either the phlebotomy or ECG area. At the phlebotomy area, technicians wash their hands at the service station and retrieve supplies for blood drawing. After venipuncture is done, staff either escort patients back to the reception desk or lead them to the ECG examination rooms if they require ECG testing. For patients who require ECG only, patients are escorted from reception to ECG directly. When staff are on break, the majority of them circulate clockwise through the clinic to gather their personal belongings from their lockers and/or to wash their hands before heading to the staff lunchroom. Although access to the staff lunchroom is closer to reception, the staff lockers, service station, and washrooms are located near the phlebotomy area. It was observed that travel to and from the staff lunchroom and lockers added unnecessary traffic in the corridors leading to the phlebotomy and ECG stations. Since these corridors are used by patients and on duty staff, they tend to be most congested around lunch time (11:30-11am). Travel patterns were used to estimate flow between functional units. The six functional units used to calculate travel distance in Section were transcribed in the row and columns of the flow matrix listed in the same order. Trips from one point and back were not always identical. For this reason, the travel matrix is not symmetric, however, the total number of times staff 46

55 entered the clinic is equivalent to the total number of staff exiting the clinic. Staff flow estimates between functional units are consolidated in Table 9: Table 9: Estimated Weekly Flow Volume between Functional Units FLOW Functional Area A B C D E F A: Laboratory Dispatch A B: Phlebotomy Stations B C: Reception C D: ECG Rooms D E: Staff Room E F: Lockers F A number of simplifying assumptions were made to derive these estimates. Access to laboratory dispatch was assumed to be strictly for dropping off blood samples from the phlebotomy area. To estimate the number of times the dispatch area was accessed, the total weekly procedure volumes for phlebotomy patients was determined and further divided by four, assuming that samples were dropped off every four procedures, as had been the observed norm. Flow in and out of the clinic was preserved. Staff were assumed to either enter the clinic from reception to phlebotomy area (trips equal to the total number of patients requiring venipuncture each week) or from reception to the ECG area (total number of ECG only patients per week). This totaled an average of 1021 staff entries into the clinic each week. The number of times staff exited the clinic area from phlebotomy to reception was calculated to be the total number of blood-only patients per week, plus half the number of blood and ECG patients per week. More patients went from reception to phlebotomy than from phlebotomy to reception, as patients requiring both bloodwork and ECG are required to have their bloodwork completed first. However, it is not necessary for these same patients to return back to reception through the phlebotomy area. The number of times staff that exited out of the clinic area from ECG to reception was calculated to be the total number of ECG only patients per week plus half the total number of blood and ECG patients per week. Again, the total number of times staff that exited the clinic area equaled 1021, which is equivalent to the number of staff that entered the clinic. 47

56 It was assumed that every time the staff went on break, the lockers and staff room were accessed from the reception, phlebotomy, or ECG areas uniformly. It was also assumed that after breaks, workers would return to the reception, phlebotomy and ECG areas uniformly as well. This explains why the number of total times staff and lockers were accessed is the same and that the section of the matrix occupied by the E and F rows and columns is symmetric. A summary of calculations can be found below: Functional Pair Lab Dispatch to Phlebotomy Phlebotomy to Lab Dispatch Phlebotomy to Reception Phlebotomy to ECG Reception to Phlebotomy Reception to ECG ECG to Reception Phlebotomy to Staff Room Staff Room to Phlebotomy Reception to Staff Room Staff Room to Reception Reception to Lockers Lockers to Reception ECG to Staff Room Staff Room to ECG ECG to Lockers Lockers to ECG Phlebotomy to Lockers Lockers to Phlebotomy Calculation (total blood + total bloodecg) * 0.25 total blood + (0.5* total bloodecg) total blood total blood only + total bloodecg total ECG only total ECG only + (0.5 * total bloodecg) total staff breaks * Activity Analysis The flow analysis related functional units on a quantitative basis. The activity analysis, on the other hand, relates functional units from a qualitative closeness rating perspective. A closeness rating scale scored from 1 to 5 shows the degree of desirability of having two functional units close together. A rating of 1 depicts undesirability, a rating of three indifferent, and a rating of 5 as being absolutely necessary. Ratings of 2 and 4 fall somewhere in between. Each of the ratings is given using a reason coded from x to z. Reasons like privacy, flow of materials, contact 48

57 necessary, and convenience, while not exhaustive, give a sense of why ratings were chosen. Pairs rated indifferent (with a 3) were not given reason codes. Table 10 shows the resulting interrelationship matrix and corresponding legends. Table 10: Interrelationship Matrix Activity Analysis Functional Area A: Lab Dispatch B: Phlebotomy C: Reception D: ECG Rooms E: Staff Room F: Lockers A: Lab Dispatch 4 x B: Phlebotomy 4 y 4 y 1 w 1 w C: Reception 4 z 3 3 D: ECG Rooms E: Staff Room 1 w 1 w 4 z F: Lockers Closeness Rating Rating Definition Absolutely necessary 5 Important 4 Indifferent 3 Unimportant 2 Undesirable 1 Reason Code Privacy w Flow of materials x Contact necessary y Convenience z Flow and Activity Relationship By comparing the functional distance matrix calculated in Section with the interrelationship matrix created in Section 5.2.4, opportunities for improvement were found, especially with 49

58 respect to the division of staff and patient space, and reducing distances between frequented areas. Table 11 below identifies candidates for improvement in orange and yellow: Table 11: Reducing and Increasing Space between Functional Areas Rectilinear Distance Functional Area A B C D E F A: Laboratory Dispatch B: Phlebotomy Stations C: Reception D: ECG Rooms E: Staff Room 26.5 F: Lockers Functional unit pairs that should be closer together to yield more efficient movement are highlighted in orange. Pairs that should be farther apart due to privacy concerns are highlighted in yellow. In terms of priority, yellow pairs should be addressed first because they are causes of worker and patient dissatisfaction and are rated 1. The orange pairs, on the other hand are optimization opportunities that are not absolutely necessary. These are rated 4 not 5 on the closeness scale Space Requirements for Redesign Based on the problem definition and as-is modeling above, space availability and requirements were determined. A summary of these findings can be found below: Space Available i. Laboratory Dispatch Area G-413 ii. Preadmit Utility Area G-420 iii. Preadmit Office G-421 iv. ECG Room G-422 Space Required i. Staff Lunchroom ii. Staff Lockers iii. ECG Waiting Area iv. Pneumatic Tube System v. Porter Cart Storage (optional) vi. Floater Room (optional) 50

59 Since it was determined that the current temporary space for the staff lunchroom is unacceptable, it is assumed that the third ECG room will be available for other use. An interview with a representative from Swiss Logic was conducted to determine the space and requirements associated with the pneumatic tube system. A designated area for the station was already determined with an associated price estimate. (Swiss Logic, 2007). However, the laboratory director suggested that the improved layout should contain the planned location of the station while the aggressive layout should place the station at the optimal location, whether that is the proposed location or not. In terms of space requirement, the base of the station is 3 feet wide and 18 inches deep. The station can be installed against a wall or free standing anywhere in the clinic. To justify an additional floater room, further analysis was conducted to determine the impact of adding an extra phlebotomy station and ECG room to the clinic without additional staff. Running a simulation over 10 weeks and changing one full time phlebotomist to perform either phlebotomy or ECG on demand, showed that the average time in the system for different patient types improved slightly depending on the day of the week. The largest impact on time in system occurs on Wednesdays, particularly during peak ECG hours. Figure 19 depicts this comparison between the baseline as-is scenario and the extra phlebotomy station and extra ECG rooms scenarios. The exact data measurements are located in Appendix D. 51

60 Wednesday - Average Time in System Average Wait (min) Base Case Xtra ECG Xtra Phleb Hour of Day Figure 19: Comparing Average Time in System on Wednesday Whether an average reduction of five minutes for ECG patients on Wednesday justifies adding an extra ECG or floater room depends on the discretion of the stakeholders. Adding an extra floater room also reduced the time in system slightly for phlebotomy patients on Monday and Tuesday but less than for ECG patients on Wednesday (Appendix D). Figures 20 and 21 show these differences in terms of average total system time from the current setup: 52

61 Monday - Average Time in System Average Wait (min) Base Case Xtra ECG Xtra Phleb Hour of Day Figure 20: Comparing Average Time in System on Monday Tuesday - Average Time in System Average Wait (min) Base Case Xtra ECG Xtra Phleb Hour of Day Figure 21: Comparing Average Time in System on Tuesdays 53

62 From a human factors perspective, workers coming into the clinic after 11am may be more motivated to work if an extra phlebotomy station or ECG room were added to the clinic. While the number of workers (11) would still outstrip the number of available workstations and examination rooms, it may provide more incentive for staff to come back from morning sweeps on time. 54

63 6 Solutions and Recommendations 6.1 Scheduling and Workflow The following three simulations were created to test their effectiveness at reducing patient time in the system, particularly prior to 10:30 and after 15:30 from Monday to Thursday. These times have previously been identified as the most problematic regions in a typical volume week (see Figure 13). Each trial was conducted twice; once using the average patient arrival rate, and once using the highest 90% arrival rates. Since all scenarios are designed to be viable options, it is outside the scope to consider cost-benefit analysis. Rather, this is simply an experiment to predict which will be most effective given the current environment Scenario 1: All Staff May Perform Both Phlebotomy and ECG Once again, the simulation was run across 10 weeks, with the average time in system (excluding medication imposed waits) collected by hour and day of arrival at the clinic (Appendix B). Figure 22 shows the generated data for average arrival rates. 55

64 Average Time in System - Scenario 1, Mean Arrival Rate Duration (minutes) Monday Tuesday Wednesday Thursday Friday Hour of the day Figure 22: Average Time in System, Scenario 1, Mean Arrival Rates The key areas of interest are those identified as having the largest wait times, namely between 7:00 and 10:00, or after 15:00, Monday through Thursday (data collected has been rounded according to the full hour). In order to compare with the basecase, consider the average profile for days Monday through Thursday, as shown in Figure 23. Friday s data has been omitted since it does not have the same wait time issues. 56

65 Average Time in System - Basecase vs Scenario 1, Mean Arrival Rate Duration (minutes) Basecase S1 All Dual Hour of the Day Figure 23: Average Hourly Time in System Comparison, Basecase and Scenario 1, Mean Arrival Rates Scenario 1 appears to have a slight advantage over the basecase. This effect is greater in the earlier part of the day, but practically vanishes after 14:00. This is to be expected as after 15:30 the only two staff on hand already perform both duties. In fact, it would be expected that the times from the two cases converge after 15:30. The fact that this does not occur at 16:00 can be attributed to either having fewer people in the queue when the shift switches over, and/or random variance in the simulation. A similar pattern is carried out for the heavier arrival rate, as shown in Figure 24. Both the basecase and sample scenario were carried out with the same elevated arrival rate, with data averaged for Monday through Thursday. The data is found in tabular form in Appendix C. 57

66 Average Time in System - Basecase vs Scenario 1, Heavy Arrival Rate Duration (minutes) Base Case S1 All Dual Hour of the Day Figure 24: Average Hourly Time in System Comparison, Basecase and Scenario 1, Heavy Arrival Rates The general shape is nearly identical, though times have been scaled up from anywhere between 1 and 7 minutes, with a greater effect at the two peaks. Under average arrival rates, patients each spend approximately 4.7 fewer minutes in the system between the hours of 8:00 and 10:30 when all technicians perform phlebotomy and ECGs. With heavy arrival rates, the drop is 4.8 minutes. While the peak between 8:00 and 9:00 appears greater than this, the higher volume of patients in the 10:00 hour skew the average. Intuitively, it should be expected that allowing staff to respond to whichever area is most in need of attention would improve patient time in system, as it would eliminate instances where one queue is empty and the corresponding workers idle, while another queue is bottlenecked. However, this simulation runs with the assumption that each technician is optimally assigned to the next patient, which may not be the case in the real world. For example, a technician could accept a patient in the wait room for an ECG, but by the time they ve traveled to the ECG facility, it may have been in use by another technician. For that reason, these results must be 58

67 handled carefully, especially when comparing them with other alternatives which model reality more closely Scenario 2: Preadmit Technician Assists with Outpatients between Appointments In order to simulate an additional worker dividing their time between preadmit and outpatients, the model was altered according to the outline in section 4.5. The new distribution of average patient time in system is given in Figure 25, with the actual data points given in Appendix B for average arrival rates, and Appendix C for heavy arrival rates. Average Time in System - Sceanrio 2, Mean Arrival Rate Duration (minutes) Monday Tuesday Wednesday Thursday Friday Hour of the day Figure 25: Average Time in System, Scenario 2, Mean Arrival Rates 59

68 Since the preadmit technician s shift is from 7:30 to 15:30, any improvement to average time in system should end after 15:00. Comparing the average daily profile for Monday to Thursday with the basecase shows the extent of the improvement. Figure 26 shows this comparison. Average Time in System - Basecase vs Scenario 2, Mean Arrival Rate Duration (minutes) Basecase S2 Pre Admit Hour of the Day Figure 26: Average Hourly Time in System, Basecase and Scenario 2, Mean Arrival Rates As expected, the effect becomes negligible after 15:00. The small disparity is likely the result of having cleared more patients from the system before the preadmit shift ends, or simple uncontrolled variation in the simulation. The greatest effect is from the hours of 7:00 to 10:00. After that point, it s likely that enough staff are on hand to deal with incoming patients without the need to form long queues, meaning there is less net improvement form having the additional help of the preadmit technician. 60

69 For heavy arrival rates (Appendix C), Figure 27 shows a similar trend is observed between the second alternative and the basecase; that is, a flattening of the spike in the morning, and little difference beyond that. Average Time in System - Basecase vs Scenario 2, Heavy Arrival Rate Duration (minutes) Base Case S2 Pre Admit Hour of the Day Figure 27: Average Hourly Time in System Comparison, Basecase and Scenario 2, Heavy Arrival Rates For this scenario, among all patients arriving between 8:00 and 10:30, the average decreased time in system was 4.9 minutes under average, and 6.1 minutes under heavy arrival rates. As with scenario 1, this alternative only assists in dealing with the morning rush period, not the afternoon s. Additionally, this model assumes that the technician will know immediately when a preadmit patient is ready for their tests, even is they are on the other side of the clinic. In order to match this to the real world, the technician will have to have a daily schedule available to them in advance while dealing with outpatients. 61

70 6.1.3 Scenario 3: Additional Worker at Times of High Demand In this scenario, an extra technician was added from 7:30 to 10:30, and 15:30 to 17:00, as outlined in section 4.5. The new weekly patient time profile, as well as Monday-Thursday comparison with the basecase, are shown below rates in Figures 28 and 29 respectively, for average patient arrival. Appendix B shows the data points. Average Time in System - Scenario 3, Mean Arrival Rate Duration (minutes) Monday Tuesday Wednesday Thursday Friday Hour of the day Figure 28: Average Time in System, Scenario 3, Mean Arrival Rates 62

71 Average Time in System - Basecase vs Scenario 3, Mean Arrival Rate Duration (minutes) Basecase S3 Extra Peak Hours Worker Hour of the Day Figure 29: Average Hourly Time in System Comparison, Basecase and Scenario 3, Mean Arrival Rate Unlike Scenarios 1 and 2, this iteration shows a distinct improvement during both the morning and afternoon peaks, making it a much more viable potential solution. During heavy arrival rates a similar pattern is repeated as shown in Figure

72 Average Time in System - Basecase vs Scenario 3, Heavy Arrival Rate Duration (minutes) Base Case S3 Extra Peak Hours Worker Hour of the Day Figure 30: Average Hourly Time in System, Basecase and Scenario 3, Heavy Arrival Rate Interestingly, this third alternative appears to be better at coping with high patient volumes than the previous two, with an average decreased time in system of 10.5 minutes, compared to 7.8 minutes under mean arrival rates between the hours of 8:00 and 10:00. For the afternoon spike (15:00 and 16:00), patients save an average of 17.8 minutes for heavy arrival rates, compared to minutes for mean arrival rates. These high numbers emphasize just how poorly having only 2 workers after 15:30 meets the needs of patients As predicted, the scenario has little to no effect on patient times in the middle of the day, when the additional shifts are not on. Observe, however, that between 11:00 and 12:00, when the shift is off, there is still a diminishing benefit, likely from the technician reducing the size of the queue while they were working. In terms of implementation, these short shifts could be accomplished by having a technician working at the core lab sent down to assist during these times (which sometimes occurs 64

73 informally to deal wit a particularly heavy volume day), or by having the OPSN staff member assigned to the clinic prior to 10:00, and a core lab member after 3:30pm Comparative Analysis between Scenarios Below are graphic comparisons between average patient time in system for the basecase and the 3 scenarios already tested (Monday to Thursday only). Figures 31 and 32, are for mean and heavy arrival rates, respectively. Average Patient Time in System, by Scenario for Mean Arrival Rates Duration (minutes) BaseCase S1 All Dual S2 Pre Admit S3 Extra Peak Hours Worker Hour of the Day Figure 31: Average Hourly Time in System, Basecase and all Scenarios, Mean Arrival Rate 65

74 Average Time in System - by Scenario, Heavy Arrival Rate Duration (minutes) BaseCase S1 All Dual S2 Pre Admit S3 Extra Peak Hours Worker Hour of the Day Figure 32: Average Hourly Time in System, Basecase and all Scenarios, Heavy Arrival Rate Both graphs show similar results: Scenario 1 and 2 have nearly identical outcomes, although having the preadmit technician help has a slightly better effect before 10:00, while having all technicians perform both function wins out between 11:00 and 13:00 Both scenarios 1 and 2 have a positive effect during the first peak, but little to no effect on the second Scenario 3 has by far the best effect during the two times of highest patient waits, but hardly any effect in the middle of the day Below is a summary of the average improvement for all patients entering the system during the key peak times, under the difference scenarios and arrival rates. Results are for Monday to Thursday, and given for both mean and heavy arrival rates. 66

75 Table 12: Average Improvement of Patient Time in System Improvement over corresponding basecase (minutes) Mean arrival rate Heavy arrival rate Scenario 8:00-10:00 15:00-16:00 8:00-10:00 15:00-16:00 1) All workers perform both tasks ) Preadmit technician assists with outpatients ) Additional worker during peak hours Once again, scenario 3 is shown to be, by far, the best alternative for addressing long wait times in the morning and afternoon. 6.2 Facility Layout Recommendations In redesigning the facility layout, primary functions were preserved for each area with the exception of the temporary staff room. Secondary functions were reallocated or consolidated to maximize efficiency of space. In particular, the laboratory dispatch area, preadmit utility room, preadmit office, and temporary staff lunch room were subjected to the most change and as they were the key sources of flexible space in the clinic Improved Layout Description Table 13 summarized the room specifications for the improved layout: Table 13: Room Specifications for the Improved Layout Room ID Room Dimension (sqt) G outpatient reception area 13x15 G-419, G ECG rooms 10x11 (G-419) 9x11 (G-420) G phlebotomy stations length 27 ft height 17 ft G-421/G staff room 10x21ft G ECG wait room 10x11 ft G joint laboratory dispatch and preadmit utility area n/a 1 staff change room 5x9 length 27ft height 17 ft 67

76 A few changes were made to the current baseline layout to create the improved one. G-418, a current ECG room, was converted to an ECG waiting area. G-420, the current location of the temporary staff room, was converted back into an ECG room. G-422 and G-421 were consolidated to form the new staff lunchroom. Staff lockers were moved to the new staff lunchroom from the crowded hallway. G-422 preadmit functions were consolidated into the G- 413 Laboratory Dispatch area where similar functions of storage and utility were also being carried out. The pneumatic tube system was placed at the edge of the laboratory dispatch area near drop off counter of the adjoining phlebotomy area. The preadmit office was eliminated due to shortage of space. It is assumed that these functions will be relocated on the preadmit side of the clinic or upstairs in the office area. Figure 33 shows the two rooms to be merged into the new staff lunchroom: Figure 33: Preadmit Office and Utility Storage Room The improved layout preserved many functions of the baseline setup. G-419 remained as an ECG room. The phlebotomy area was left untouched with seven workstations. Both the outpatient reception and the staff change room were unchanged. While the dimensions of the laboratory 68

77 dispatch space were preserved, counters were rearranged or removed to better accommodate preadmit utility functions and the porter carts. Figure 34 shows the floorplan of the improved layout. (Pneumatic (Porter Carts) Tube) G-413 Laboratory Dispatch / Utility Area G-417 Phlebotomy Area G-418 ECG Wait Area G-414 Outpatient Reception G-419 ECG Room Staff Change Room G-420 ECG Room G-400 Shared Wait Area (Lockers) G-421 / G-422 Staff Room Figure 34: Improved Layout 69

78 6.2.2 Evaluation of Improved Layout In the combined flow and activity analysis completed in the as-is situation, a number of opportunities for improvement were found. High priority items included moving the lockers and staff room closer together to consolidate staff space, and separating patient and staff space for privacy reasons. Lower priority improvement areas associated with increasing efficiency between the laboratory dispatch area and phlebotomy, as well as phlebotomy to ECG, were also identified. The improved layout addresses most of these concerns. The staff room and lockers were consolidated into one space, away from the work area, and made accessible through their own hallway. The distances from the laboratory dispatch area to phlebotomy was preserved as no structural changes were made to these areas. However, the distance from phlebotomy to the closest ECG room was lengthened, due to the conversion of ECG room G-418 into an ECG wait area and temporary staff room G-420 back to an ECG room. Table 14 shows more general ways for the improve layout to be evaluated, besides meeting location constraint and preferences. Table 14: Other Criteria for Evaluating the Improved Layout Other Criteria Ease of implementation Comments Only one structural change related to the staff lunchroom was suggested. Buy-in from preadmit staff will be required in order for construction to go ahead. Otherwise, implementation should be relatively straight forward as furniture can be easily removed or rearranged. Flexibility of layout Space utilization Employee satisfaction Some flexibility is offered in this layout with regards to preadmit utility functions being consolidated into the laboratory dispatch area. How this should be implemented is up to the discretion of the outpatient and preadmit supervisors. Space utilization is moderately efficient in this layout. The third ECG room converted into a wait area for ECG is efficient use of space if there are not enough resources to justify a third ECG room. Expected to be high. Preliminary drawings were reviewed informally by staff. Staff were extremely happy with the prospect of a new staff room with enough space to contain their lockers and that the phlebotomy area was left unchanged. Staff requested for implementation as soon as possible. 70

79 6.3.1 Aggressive Layout Description Table 15 summarized the room specifications for the aggressive layout: Table 15: Room Specifications for the Aggressive Layout Room ID Room Dimensions (sqt) G outpatient reception area 13x15 G-419, G ECG rooms 10x11 (G-419) 9x11 (G-420) G dual ECG/Phlebotomy room 10x11 G preadmit utility area 9x10 G staff room 10x21 G phlebotomy stations length 36 ft height 17 ft G laboratory dispatch area length 36 ft height 17 ft phlebotomy corridor 1 ECG Wait area 3x10 n/a 1 staff change room 5x9 A number of changes were made to the baseline layout to create the aggressive layout. As suggested in the improved case, the old preadmit area and office were merged to form the new staff room, and lockers were subsequently moved there. To replace the old preadmit utility room, a new location was offered, adjacent to the staff change room. Figure 35 shows the area suggested for conversion into the new preadmit utility room: Figure 35: Space for New Preadmit Utility Room 71

80 To accommodate this new room, it was suggested that a preadmit desk be removed. During approximately ten visits to the hospital over a period of six months, a preadmit desk was always observed to be vacant. Figure 36 shows the vacant desk during a recent visit. Figure 36: Vacant Desk in Preadmit Area Like in the improved layout, the temporary staff room G-420 was converted back to an ECG room. Unlike the improved layout, it was suggested that room G-418 ECG be converted into a dual ECG/phlebotomy room. Figure 37 shows that there already exists a blood draw chair in room G-418 to accommodate phlebotomy. Figure 37: Blood draw Chair in ECG Room 72

81 Major changes were proposed for the phlebotomy and laboratory dispatch areas. The phlebotomy area was expanded into the laboratory dispatch area. Walls were added in the phlebotomy corridor to alleviate patient privacy concerns and also to accommodate the ECG wait area. An extra phlebotomy station was added to the area bringing the total number of workstations to eight. Six stations were arranged along the far wall, and the service station was re-centered. The pneumatic tube station was moved from the laboratory dispatch area to a dedicated corner of the phlebotomy area. In the laboratory dispatch area, the specimen dropoff window, shared with the adjoining phlebotomy area, was removed and replaced with a door. Like in the improved layout, counters were rearranged or moved in the dispatch area to accommodate the porter carts. The middle G-419 ECG room, reception space, and staff change room were unchanged in the aggressive layout. Figure 38 show the reception area and staff change room preserved in their current state. Figure 38: Outpatient Reception Area and Staff Change Room Figure 39 on the next page shows the floorplan of the aggressive layout: 73