SIMUL8 User Group - Geisinger Case Study - Balancing Patient Satisfaction with Staff Efficiency
1 like793 views
The document discusses an initiative by Geisinger Medical Center to transition its food service department to an on-demand room service style delivery system for inpatients, outlining objectives, modeling methodologies, and outcomes of the pilot phase. Key findings include a trade-off between patient service and resource efficiency, with recommendations for optimal cart capacity and dispatch times. The ongoing phase two modeling aims to enhance service levels through improved staff scheduling and delivery processes.
SIMUL8 User Group - Geisinger Case Study - Balancing Patient Satisfaction with Staff Efficiency
- 1. 2013 East Coast User Group Meeting Balancing Patient Satisfaction and Staff Efficiency: Logistics of Meal Delivery December 11, 2013
- 2. Presenter Seth Hostetler Process Engineer, Care Support Services Geisinger Health System • Care Support Services includes Supply Chain and Enterprise Pharmacy • Been with Geisinger for 2+ years • Educational background is in Industrial Engineering and Operations Research • Member of IIE, SHS, and INFORMS SIMUL8 Corporation | SIMUL8.com | info@SIMUL8.com 1 800 547 6024 | +44 141 552 6888
- 3. Agenda • Introduction and objectives • Approach and the proposed system • Phase 1 Modeling • Experiment • Results and insights • Recommendations and future directions • Phase 2 Modeling • • • • Model adjustments Operational strategy testing Analysis and recommendations Complete experimental results |3 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 4. Introduction • Geisinger Medical Center’s (GMC) food service department was planning a transition to on-demand, room service style delivery of meals to inpatients in 2013 • 18 inpatient units, approx. average census = 360 • Excludes inpatient psych unit • The new system will result in: • Changes in resource management and requirements • Changes in process and employee roles • Changes in patient service |4 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 5. Objectives 1. Create a representation of the new food service delivery process in a simulation environment 2. Use simulation to show how changing system parameters will affect patient service levels and resource requirements 3. Through model analysis, create operational recommendations for the forthcoming system changes |5 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 6. The Model and the Team • This work is based on a model that we have been using building and working with since 2010 • Original model was developed by a consulting group, Efficiency Engineers • Geisinger did not have the in-house expertise to create • Able to leverage consulting expertise to develop the complex model • Helped to gain buy-in from senior leadership to help develop the process engineering team • Model was handed off to the engineering to for further development and use • Phase 1 of this work was the thesis research of an Industrial Engineering Masters student at Penn State Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary |6
- 7. The Internal Logistics Simulation Model • Captures all support service functions within GMC • 19 inpatient units • 33 outpatient clinics • 51 ancillary service departments • Captures food delivery to: • Med/Surg. Units • Adult critical care units • Women’s and Children’s |7 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 8. Internal Logistics Simulation Model 2.1 Miles of Hallway 38 Elevators 11 Floors 199 Network Nodes 160 Destinations 35 Transport Groups Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 9. Modeling the New Food Services System • Three delivery zones • Dispatch timer on carts • Once the first tray is placed on a cart a timer is set • The cart will leave when it is full or when the timer expires, whichever occurs first • Nearest neighbor delivery route • Cart and staff resource modeled as one entity Zone A Zone B Zone JK AGP4 AICU4 SCU4 AGP5 ICS5 AICU5 GP2 SCU3 SCU5 BP5 BP6 BP7 BP8 WILL1 CH2 CH3 HfAM7 HfAM8 • Always a host to travel with a cart • Three meals per day • Approximately 1060 meals served between 5:30am and 9:30pm |9 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 10. The Delivery Process Patient orders meal Is there a cart waiting for that zone? N Initiate new cart for the zone Y Add meal to zone cart Is cart at capacity ? N Y Has the time limit been met? N Y Wait for another meal or until time limit met Send trays for delivery | 10 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 11. Input Data • Distribution of meal orders • Estimate of meal delivery count every 30 minutes • Patient census snapshot • Time to serve meal to patient • Estimated with time studies • ~59 seconds to serve meal • ~49 additional seconds for isolation patient • 10% of patients modeled as isolation status • Travel speed • 185 feet/min | 11 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 12. Key Performance Measures PATIENT SATISFACTION • Service time OPERATIONAL EFFICIENCY • Cart utilization • Wait time on cart before routing • Maximum number of carts used* • Percent delivered after 30*, 40, 45 and 50 minutes • Number of delivery trips made in a day * Management chose these two metrics as the most important. | 12 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 13. The Phase 1 Experiment • Carts modeled as an unlimited resource • Three dispatch times • 8, 10, or 12 minutes • Three cart capacities • 12, 14 or 18 trays Which combination of dispatch time and cart capacities would be most effective? | 13 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 14. Results Dispatch Timer Cart Utilization Service Time Max # Carts # Delivery Trips Service Level Parameters Minutes Cart Capacity 8 12 8 14 8 18 10 12 10 14 10 18 12 12 12 14 Parameters Minutes Cart Capacity 8 12 8 14 8 18 10 12 10 14 10 18 12 12 12 14 Service Time Average 17.22 17.20 17.23 19.23 19.31 19.31 20.97 21.20 95% CI Average 95% CI Average 95% CI (17.15, 17.29) 16.00 (15.83, 16.17) 242.60 (241.40, 243.80) (17.13, 17.27) 15.93 (15.66, 16.21) 242.70 (241.39, 244.01) (17.17, 17.29) 15.90 (15.72, 16.08) 243.23 (241.64, 244.82) (19.16, 19.30) 14.63 (14.45, 14.82) 206.53 (205.36, 207.70) (19.23, 19.38) 14.64 (14.46, 14.83) 205.00 (203.77, 206.23) (19.23, 19.39) 14.37 (14.18, 14.55) 205.43 (204.39, 206.47) (20.89, 21.04) 13.70 (13.53, 13.87) 179.60 (178.82, 180.38) (21.11, 21.30) 13.27 (13.10, 13.43) 177.77 (176.79, 178.75) Cart Utilization Average 0.3613 0.3086 0.2399 0.4235 0.3660 0.2850 0.4878 0.4238 Max Number of Carts Number of Delivery Trips 95% CI (0.3597, 0.3629) (0.3068, 0.3105) (0.2387, 0.2412) (0.4210, 0.4259) (0.3640, 0.3681) (0.2835, 0.2864) (0.4856, 0.4899) (0.4217, 0.4260) Wait Time on Cart Before Routing Average 95% CI 4.78 (4.77, 4.80) 4.80 (4.78, 4.83) 4.80 (4.78, 4.81) 5.80 (5.82, 5.77) 5.83 (5.81, 5.85) 5.85 (5.83, 5.87) 6.72 (6.69, 6.76) 6.81 (6.78, 6.84) Percent delivered after 30 min Average 95% CI 2.06 (1.83, 2.29) 2.00 (1.78, 2.21) 2.10 (1.92, 2.29) 5.67 (6.02, 5.32) 6.08 (5.78, 6.38) 5.97 (5.59, 6.35) 10.79 (10.41, 11.16) 12.04 (11.58, 12.49) Cart Capacity Cart Utilization Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Percent delivered after 40 min Parameters Minutes Cart Capacity Average 95% CI 8 12 0 (0, 0) 8 14 0.003 (-0.003, 0.010) 8 18 0.003 (-0.003, 0.010) 10 12 0.061 (0.026, 0.095) 10 14 0.117 (0.073, 0.161) 10 18 0.199 (0.129, 0.270) Not12 reuse or distribution without permission for 12 0.231 (0.175, 0.288) Geisinger Health System Confidential and Proprietary 12 14 0.434 (0.310, 0.558) Percent delivered after Percent delivered after 45 min 50 min Average 95% CI Average 95% CI 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0 (0, 0)| 14 0.013 (-0.003, 0.028) 0 (0, 0) 0 (0, 0) 0 (0, 0) 0.010 (-0.001, 0.020) 0 (0, 0)
- 15. Interpreting the Results • These trends reveal a broad theme within the meal delivery system: A trade-off exists between patient service and efficient use of resources. • The 8 minute dispatch timer provides the best performance in terms of percent of meals delivered beyond each of the time intervals. • As the dispatch timer increases, the results show an improvement in the efficient use of resources; however, the factors measuring patient service perform worse as the timer length grows. • A cart capacity of 12 meal trays provides the highest utilization for each of the three dispatch timer settings. – Cart utilization is the only measure significantly affected by a change in cart capacity. – For this reason, the 12 meal tray carts were selected. | 15 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 16. Results 12-Tray Cart Capacity Max Number of Carts Number of Delivery Trips Cart Utilization Parameters Minutes Average 95% CI Average 95% CI Average 95% CI 8 16.00 (15.83, 16.17) 242.60 (241.40, 243.80) 0.3613 (0.3597, 0.3629) 10 14.63 (14.45, 14.82) 206.53 (205.36, 207.70) 0.4235 (0.4210, 0.4259) 12 13.70 (13.53, 13.87) 179.60 (178.82, 180.38) 0.4878 (0.4856, 0.4899) Parameters Minutes Percent delivered after 30 min Average 8 2.06 95% CI (1.83, 2.29) % Delivered after 30 minutes 12 12 Minutes 10 Percent delivered after 40 8 min 6 Average 10 Minutes 95% CI 4 0 (0, 0) 8 Minutes 2 10 5.67 (6.02, 5.32) 0.061 (0.026, 0.095) 12 10.79 (10.41, 11.16) 0.231 14 14.5 (0.175, 0.288) 0 13.5 15 15.5 Max Number of Carts 16 16.5 | 16 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 17. Usage Profile for the 12-Tray Cart 18.00% Percent of Time in Use 16.00% 14.00% 12.00% 10.00% 8.00% 8 Minutes 6.00% 10 Minutes 4.00% 12 Minutes 2.00% 0.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Number of Carts in Use • Must consider more than just the max number of carts used. • For example, the 8 minute dispatch timer experiences a count of 17 carts in use for a period of time in a day, but it is for less than 0.01% of the time. • A balance must be struck between the need to have a cart available to deliver meals, and the acceptable limit of time that a meal must wait to be delivered because it is waiting for a cart. | 17 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 18. Phase 1 Recommendations • 12 tray cart capacity will provide best cart utilization and lowest equipment cost • Depending on dispatch timer setting, trade-offs exist between: • • • • Required number of carts (14-17) Number of delivery trips (178-244) Cart Utilization (35%-49%) Service Level (1.83%-11.16% delivered after 30 min) • Shorter timer will provide better service level • Longer timer will sacrifice service level for improved resource usage | 18 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 19. Phase 2 Model • The operational staff schedule provided did not provide enough employee hours to satisfy work requirements • Utilization is greater than 100% • Combined delivery and pickup processes determined to enhance service levels • Estimated approx. 40% reduction in number of trips required • Inclusion of return process • Empty cart swapped with cart full of dirty trays on return route • Return trip route based on a cyclical zone schedule • Carts cleaned after each round trip | 19 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 20. Additional Model Assumptions • Food preparation and cart cleaning time included • Zone routing for pickup and delivery • Carts modeled as an unlimited resource • Cart use limited by number of staff available • No break times included in resource schedule • At request of management • Adjustments made to original order distribution to create a 7am-7pm order estimate Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary | 20
- 21. Performance Measures 1. Service level – Percent of carts delivered in 45 minutes or less 2. Resource utilization – Percent of time resources are in use for delivery/pickup services 3. Scheduled work hours in one day | 21 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 22. Experimentation Objective • Determine the optimal staffing schedule, as well as the most efficient delivery and pickup routing methods • Test various daily employee schedules to determine an hourly schedule which allows for high service levels and worker utilization • Management asked that staff be scheduled hourly, not by shift • Ability to flex staff across various tasks Example Schedule Time Staff Count 7am-8am 9 8am-9am 12 9am-10am 12 10am-11am 8 11am-12pm 8 12pm-1pm 13 1pm-2pm 12 2pm-3pm 10 3pm-4pm 9 4-pm-5pm 7 5pm-6pm 9 6pm-7pm 11 7pm-8pm 9 8pm-7am 0 | 22 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 23. Delivery and Pickup Routing Scenarios Testing performed on three scenarios: A. Standard original operations (3 zones, fixed route schedule) B. Modified pickup schedule • • Rotational delivery schedule forces delivery route to begin from different unit on each trip Pickup unit is last unit in delivery schedule C. Six zone facility layout with original operations Zone Units 1 2 3 4 5 6 AGP4 AICU4 SCU4 AGP5 AICU5 ICS5 GP02 SCU3 SCU5 BP05 BP06 BP07 BP08 WLL1 CHM2 CHM3 HFAM6 HFAM7 HFAM8 Which of these scenarios do you think will be most effective? | 23 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 24. Scenario Testing • All scenarios tested in a 24-hour simulation period • Includes routing and staffing variations • Seven highest performing scenarios tested in 30-day simulation period • Reduction of variability • Testing to maximize service level and utilization • Minimization of total employee hours used to break “ties” | 24 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 25. Comparison of Service Level (squares) and Resource Utilization (triangles) for the 4 best performing staff schedules. Option A: Shown in green, is the standard routing, 3-zone option. Option B: Shown in red, is the modified, rotational, 3-zone option. Option C: Shown in blue, is the standard routing, 6-zone option. 105 100 98 95 96 95 94 90 99 98 97 98 97 96 96 95 90.52 88.4 87.5 Standard Util 85 85 83.7 Standard Service Rotational Util 80 80.8 80 Rotational Service 79.75 6 Zone Util 6 Zone Service 75.86 75 74.83 74.78 70 69.7 65 120 129 Heal • Teach • Discover • Serve 3 7 131 142 60 Copyright Geisinger Health System 2013 Not for reuse or distribution without permission 6 Geisinger Health System Confidential and Proprietary Schedule 5 | 225 5
- 26. Six Zone Delivery Layout 98% Service Level ~20/meals per day delivered after 45 minutes 88.4 % Utilization 114 hours/day busy (129 total) 6 Zone Operation Utilization-Service Trade-Off 110 105 100 96 98 99 95 90 85 90.52 88.4 80 Utilization 83.7 Service 75 70 65 60 3 7 Schedule 5 * Schedule 6 was eliminated from comparison because it performed comparable to Schedule 7, however the utilization was slightly lower and it used additional scheduled work hours Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary Schedule 7 Time Staff Count 7am-8am 8 8am-9am 13 9am-10am 12 10am-11am 7 11am-12pm 8 12pm-1pm 13 1pm-2pm 11 2pm-3pm 10 3pm-4pm 8 4-pm-5pm 7 5pm-6pm 10 6pm-7pm 12 7pm-8pm 10 8pm-7am 0 | 26
- 27. Conclusions and Recommendations • Six zone facility layout allows for best service with least compromise in employee utilization Zone Units 1 2 3 4 5 6 AGP4 AICU4 SCU4 AGP5 AICU5 ICS5 GP02 SCU3 SCU5 BP05 BP06 BP07 BP08 WLL1 CHM2 CHM3 HFAM6 HFAM7 HFAM8 • Best found schedule requires 129 hours scheduled for delivery/pickup services per day • Break times not included in schedule | 27 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 28. Final Thoughts • We chose to use Simul8 due to the ease of use, system performance, and low-cost • The simulation model has been great at helping us gain stakeholder buy-in from various groups • Allows us to model test of change before implementing in actual system • We have used a variety of resources to help us with our model • Consultants helped with initial build and continue to help with difficult problems on an ad-hoc basis • Students have developed pieces of the model • Process engineering team provides in-house support | 28 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 29. Final Thoughts Keys to Success • Involve the process stakeholders • Good communication with the management and staff of the processes you are modeling is essential. • Take time to fully understand the system • Do the job, or shadow the job, to truly understand the processes. • Gather accurate data • Don’t always just trust system to provide good data, you may need to manually collect the data yourself. • Build and test iteratively • Build the model in small pieces and constantly test it to make sure your model is performing how you desire. | 29 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary
- 30. THANK YOU FOR ATTENDING! WHO HAS QUESTIONS? Seth Hostetler Geisinger Health System sthostetler@geisinger.edu 570-214-7029 Heal • Teach • Discover • Serve Copyright Geisinger Health System 2013 Not for reuse or distribution without permission Geisinger Health System Confidential and Proprietary | 30