Strategies to Improve Emergency
Department Patient Flow – Our
Experience
NewYork-Presbyterian Hospital
Columbia University Medical Center
May 21st, 2007
NYS DOH Patient Safety Conference
Robert Green, MD, MPH
Year 2000
• ED LOS
14 hours
(Admitted Patients)
• Left without being seen
8%
• Diversion hours
1081 annually
21 hours/week
NewYork-Presbyterian Hospital
Columbia University Medical Center
Monthly Patient Visits Jan 1997 - April 2004
5500
Patient Visits
5000
4500
4000
3500
3000
1
4
7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88
Month
Indicators of ED Throughput
Efficiency
• ED Length of Stay (Overall)
– Admitted
– Discharged
• Diversion
• Walkouts
Intrinsic versus Extrinsic causes of
Throughput Problems
• Intrinsic
– Staffing levels
– Productivity
• Extrinsic
–
–
–
–
–
Inpatient bed availability
Lab/Radiology Turnaround Time
Transport
Nursing Report to Inpatient Unit
Inpatient Physician Assignment
Key Points
• Hospital Administration: Understanding and
willingness to tackle extrinsic causes of ED
throughput issues
• Highest level involvement at regularly scheduled
multidisciplinary meetings focused on the
function of the ED
• (Extrinsic) Change does not occur or reverts
quickly back to baseline without this high level
involvement
Key Points (2)
• The case for change and new initiatives was
made with DATA
– If we can’t measure it, we can’t improve it.
• Development of novel ideas specific to your ED
• Change serial processes into parallel processes
• Optimize current staffing allocation based on
arrival patterns and queueing theory
• Series of incremental small changes resulted in
significant improvement in throughput
Multidisciplinary Team
• Led by Chief Operating Officer and VP for
Operations
• Committee composition:
– Dept of Emergency Medicine and Dept of Medicine
Senior Leadership
– Nursing Administration (ED and Hospital)
– Admitting/Census Director and Associate Director
– Housekeeping Director
– Finance
Initiatives
2001-2007
•
Intrinsic
–
–
–
–
•
Queueing Model and reallocation of physician staffing hours
Productivity feedback to physicians
Additional nursing and physician resources added
Movement of Patient To Inpatient Location after bed assigned S -> P
Extrinsic
–
–
–
–
–
–
–
–
Inpatient Hallway Bed Policy S -> P
No-delay Faxed Nursing Report
Pre-Diversion Policy
Department of Medicine Admission Assignment Policy (MAR) S -> P
Early Bed Request S -> P
Transportation Request Policy (Pending) S -> P
Bed Tracking System for Inpatient Bed Identification and Housekeeping
Transport of Patients To ICU
NewYork Presbyterian Hospital
Scatter Plot of Length of Stay and Walkouts
11.0
10.0
9.0
8.0
7.0
6.0
y = 1. 6 3 4 1x - 3 . 6 3 4 8
R 2 = 0 .4 5 7 4
5.0
4.0
3.0
2.0
4
4.5
5
5.5
6
6.5
7
Me an We e kly Le n g th of S tay
7.5
8
8.5
9
NewYork-Presbyterian Hospital
Columbia University Medical Center
Diversion Hours
1200
1000
Pre – Diversion Policy
800
Hours
600
400
200
0
1999
2000
2001
2002
Year
2003
2004
Prior Admitting System - MAR
Inpatient Admitting Teams
New System of Admitting
Inpatient Admitting Teams
Admitting Program
• Developed by ED personnel for use in new
admitting system for the Department of
Medicine
• Resulted in 1 hour decrease in overall
Length of Stay
• See Computerized Admit Board
Queueing models
•
Efficiency is often a key dimension of good service. In the ED it
is an important factor in patient satisfaction and ED throughput
•
Delays result from short-term, unpredictable fluctuations in
demand and capacity
•
Queueing phenomenon is complex and impossible to predict
without appropriate tools
•
Good performance is particularly difficult when:
– Relative amplitude (ratio of peak to average demand) is
high
– Staffing periods are long (e.g. more than 1 hour)
– Service times are long (e.g. more than ½ hour)
Monday
ED Arrival Pattern at Allen Pavilion Hospital
8
7
Arrivals/hr.
6
5
4
3
2
1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
Time
14
15
16
17
18
19
20
21
22
23
24
An Example of the Lag Phenomenon
λ = 0.2, μ = 0.2, RA = 1, s = 7
0.35
0.14
Delay Probability
0.30
0.12
Time
48
46
44
42
40
38
35
33
31
29
27
25
Arrivals pre Hour
0.00
23
0.00
21
0.02
19
0.05
17
0.04
15
0.10
13
0.06
10
0.15
8
0.08
6
0.20
4
0.10
2
0.25
0
Probability of Delay
Arrival Rate
12
:0
0
1: AM
00
2: AM
00
3: AM
00
4: AM
00
5: AM
00
6: AM
00
7: AM
00
8: AM
00
9: AM
00
10 A
:0 M
0
11 A
:0 M
0
12 A
:0 M
0
1: PM
00
2: PM
00
3: PM
00
4: PM
00
5: PM
00
6: PM
00
7: PM
00
8: PM
00
9: PM
00
10 PM
:0
0
11 P
:0 M
0
PM
P (Wq(t)>1hr)
1
0.7
0.6
2
0.5
0.4
1.5
0.3
1
0.2
0.1
0.5
0
0
Time
Staffing level
Staffing Levels and Estimated Pr (Delay > 1 hr)
Avg. service time = 45 minutes
3.5
0.9
3
0.8
2.5
Allen Pavilion ED
Results of Physician Staffing Rearrangement
Utilizing Queuing Theory Model
Oct 2002 – Oct 2003 –
Feb 2003
Feb 2004
Patients
4807
5354
treated
Patients
431 (8.2%) 412 (7.1%)
who LWBS
13% Reduction
Total
5238
5766
Inpatient Hallway Bed Policy
• Prior policy – patients wait for a clean
assigned bed
• Serial processing versus parallel
processing
• Patients can wait in the hallway next to
room if not yet ready upon arrival
• Communication with the patient – Manage
Expectations