Addressing overcrowding in a cardiovascular outpatient clinic using discrete-event simulation Vahab
1 Vahdatzad (
[email protected]),
Jacqueline Griffin PhD
1,
James Stahl MD, CM, MPH
2
1 Department of Mechanical and Industrial Engineering, Northeastern University; 2 Massachusetts General Hospital Institute for Technology Assessment
Background
3
System model
Conducted at cardiovascular outpatient clinics
• 3 clinics with 26 exam room s and over 300 clinicians • Over 45,000 visit per year - Although most of the patients are scheduled, clinics are overcrowded and patients have to wait a long time. OBJECTIVE: Build a simulation model of to gain a greater understanding of patient flow and to analyze the system efficiency from patient and hospital perspective. • Decreasing patients’ length of stay. • Increasing medical assistants efficiency and room utilizations
Patient flow legend Check-in Waiting area Examination room Phlebotomy Check-out Leave clinic
1
4
2 3
5 3
4
2 3
1
4
5
5 6
Results
5
6
• Different patterns of patient flow • Finding bottlenecks from clinic point of view • Using computer simulation to model the current state of the clinic and understand the possible solutions to overcrowding
Patient length of stay division Divisions
1
CI Waiting area
CI
MA
Physician visit
0
10
20
30
40 time
Physicians
14 Rooms
9 physician
MA availability 26%
Resources
Medical assistants (MA)
EKG machines
4MA
2
50
60
70
80
Waiting for physician 39%
Vital 20% EKG 70%
• Pooled allocation policy leads to less waiting time in order to initiate treatment
Sample patient flow when a patient needs phlebotomy test after being visited by a physician.
MA Medical Assistant
Phlebotomy
Waiting for EKG 10%
Limited resources that cause overcrowding Exam rooms
CO
Check-in
CO Check-out
Exam room availability 74%
6
Legend
Physician visit 61%
Effect of room allocation policy on waiting time 25
Total waiting time
20 15
Waiting time for examination room Waiting time for MA
10 5 0
Designated
Pooled
Hybrid
3 EKG
Data collection and model building
• Clinical reports were available for patients’ appointment times from Aug/1/2012 to Jul/31/2013 • Within two observation periods, data are collected for other processes related to patient flow. • Modified distributions are used as input to the simulation. 60min 40 min 5% 7%
15 min 13%
30 min 28% 20 min 47%
Appointment time variation and probabilities
• Developing a Discrete-event simulation model • Elaborative details are considered in building the model • - Room allocation policies • - Physicians and MAs non-patient related activities(e.g. Paper works, documentations, room preparation) are considered separately. • - Staff schedules are added to the model. • • Number of runs: 255 days
• Clinic closing condition: If all patients have left the system and no paperwork/administration tasks are left for physicians or MAs.
4
Model validation
• Simulation model was first validated by clinic administrative staff • We also verified the simulation results using t-test • No significant difference between actual data and simulation results(α = 0.05)
• Simulation results help administrators to predict changing resource levels costs and benefits tradeoff. • Adding one physician decrease length of stay significantly Impact of changes to resource levels on length of stay 78 73 Length of Stay
68 63
80
Current LoS
Comparison of simulation output with collected data
70
6
60 50 40 Actual data Simulation output
30 20 10 0 Waiting time before treatment initiation
Waiting for physician
Patient Legnth of Stay
Adding EKG machine
Adding MA
Adding Physician
Future research
• Develop real-time location system tools - Improve patient care - Decrease patient length of stay - Increase resource utilization