Designing Patient Flow in the Hospital

to Make Patients Safer

John B. Chessare, MD, MPHInterim President, Caritas Christi Health Care System

President, Caritas Norwood HospitalSenior Vice President for Quality and Patient Safety,

Caritas Christi Health Care System

Why should we care about patient flow?

1. To make our patients safer

2. To increase throughput (volume, $$)

3. To reduce expenses (cost, $$)

4. To improve staff satisfaction

5. To improve patient satisfaction

A question to run on …….

What can I do as a healthcare leader to improve patient flow?

Agenda

• Introduction 5 minutes– What is the fundamental problem? – What management model will help us improve it?

• Some examples of designing flow 15 minutes– Smoothing Flow at Boston Medical Center:

Changing the Surgical Schedule– Designing Flow out of the Emergency Department at Caritas

Norwood Hospital

•US Air 562 from Boston to Albany in its final approach

•Captain: “Albany this is US Air 562”

•Air Traffic Controller: “Roger US Air 562 this is Albany Control. You’llhave to hold at your present altitude. We’ve got a lot more planes in ourairspace than usual. The airlines decided to add some flights but no onetold us and we’ve got some rerouted planes due to bad weather in metro New York.”

Luckily, this type of communication does not happen in commercial aviation…….

•US Air 562 from Boston to Albany in its final approach

•Co-pilot: “Boy, we’ve got to get this plane down or we’ll havesome angry passengers. There’s the airport. Lets pick a runway.I usually call the gates myself and find out if any are open andthen I just go for it. If you don’t, the controller will give it tosomeone else”

•A Physician and Two Nurses Discussing a Patient in the EDWaiting to Be Admitted

•Physician: “ This guy is ready to go upstairs. Its now 5pm, hecame in at 10 this morning. The unit clerk called admittingbut I guess they are at dinner”.

•First Nurse: “Ok, I’ll call around to the floors and see if thereare any empty beds….I know who to call.”

•Second Nurse: “Oh, I usually call the supervisor. Did you callreport?”

•First Nurse: “Oh no, I leave it on the floor’s voicemail just before I leave the ED with the patient so they can’t slow the transfer down”.

Hospitals have been managed sub-optimally

• Too much is happening by chance. Too little is happening by design and therefore we function at low reliability

• Managers have been managing inputs: studies per FTE; deviation from budget, etc. but not the system.

• The hospital is full of batching; Patients are admitted and discharged in batches. Tests are run in batches. Surgeries are done in batches without consideration of the effect on the system.

• Safe patient care is easier to reach with continuous flowand not with the artificial variability of batching!

• There is a need for scientific management in the hospital industry

Reason’s Swiss Cheese Model of Error

Defence-in-depthUnsafe acts

Psychologicalprecursors

Latent failures at themanagerial levels

Local triggersIntrinsic defects

Atypical conditions

Trajectory ofaccident opportunity

We allow patients to aggregateand move in batches that overwhelm our staff

We are managing the efficiencyOf individual inputs and not thesystem

“Hard work and good intentions are necessary but insufficient for exceptional care”.

“Every System is perfectly designed to get exactly the results that it gets.”

Time

# of

Pat

ient

s

Variability

1. “Natural”: you can’t control it …you just have to manage it.(e.g.. sick patients coming to the ED). Tool to manage it: queuing theory

2. “Artificial”: you can control it….you must eliminate it to create flow. (batching) (e.g. elective surgery scheduling, reading stress tests)

When we “batch and push” we create artificial peak loads that create overcrowding• Internal Diversion –patients sent to alternative

floors\Intensive Care locations• Internal Delays – PACU backs up• External Diversion - ED diversion; inability to accept

transfers• Staff overload – increased errors and staff unhappiness• System Gridlock – Increase in LOS• Decreased Volume• Unhappy patients

What business model should we use to improve flow?Performance Improvement

1. Focus on the patient and his or her family2. Deep Process knowledge (Design)3. Decisions driven by data4. Teamwork5. Empowerment

“How can we use the ideas of individuals on the team to redesign our systems to measurably improve the health and satisfaction of our patients and their families while driving out waste?”

Flow Teams at Boston Medical Center

Flow LeadershipTeam

ED Team Inpatient Team Surgical Scheduling Team

Average total ED throughput timeBoston Medical Center

0

1

2

3

4

5

61 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69

Weeks

Hou

rs

Series2 Series3

Improvement from 4.5 to 3.75 hours30 minutes x 1050 cases = 31,500 minutes or 525 hours per week saved

Improving Inpatient FlowTeam

• Janet Gorman• John Chessare• Linda Guy• Jane Damata• Dina Brauneis• Brian Brisbois

• Sue Doherty• Jacque O’Shea• Cil Weekes• David Roney• Kim Wood

The Inpatient Cycle, Key Points, Key Process IndicatorsClean Bed

Bed Assigned

PatientArrivesIn Bed

Bed Assignment to Arrival Time

Patient Clinically Ready to Leave

Length of Stay

PatientLeaves

Average Discharge Time

Bed Turnover Time

Dirty Bed

120 minutes

5.5 days

15:01

60 minutes

Maximizing Throughput:Smoothing the Elective Surgery Schedule to Improve Patient Flow

James M. Becker, MDKeith P. Lewis, MDJohn B. Chessare, MD, MPHEugene Litvak, PhD

Richard J. Shemin, MDGail Spinale, RNDemetra OuelletteAbbot Cooper

Surgical Smoothing

1. Smoothing Elective Vascular Surgery2. Smoothing Elective Cardiac Surgery3. Separating Elective From Urgent Surgery in the Menino

Pavilion• Creating reliable urgency data• Separating a room for urgent/emergent cases• Eliminating Block Scheduling

0

0.2

0.4

0.6

0.8

1

1.2

Mon

7W

Tue

Wed

Thu

Fri

Mon

8W

Tue

Wed

Thu

Fri

Mon

PC

U

Tue

Wed

Thu

Fri

Sat

Mon

SIC

U

Tue

Wed

Thu

AbramsonRestucciMadisonSampsonWong

Bed Need by Day of Week for Vascular Surgery (18 months of data)

Progressive Care Unit

Volume

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

7/1/20

027/2

/2002

7/3/20

027/4

/2002

7/5/20

027/6

/2002

7/7/20

027/8

/2002

7/9/20

027/1

0/2002

7/11/2

0027/1

2/2002

7/13/2

0027/1

4/2002

7/15/2

0027/1

6/2002

7/17/2

0027/1

8/2002

7/19/2

0027/2

0/2002

7/21/2

0027/2

2/2002

7/23/2

0027/2

4/2002

7/25/2

0027/2

6/2002

7/27/2

0027/2

8/2002

7/29/2

0027/3

0/2002

7/31/2

002

Volume

Vascular Elective PCU Cases by DayRandom Month July 2002

Vascular Scheduled PCU Cases - Weekdays Only (October 2003)

0

1

2

3

4

5

10/01

/0310

/02/03

10/03

/0310

/06/03

10/07

/0310

/08/03

10/09

/0310

/10/03

10/13

/0310

/14/03

10/15

/0310

/16/03

10/17

/0310

/20/03

10/21

/0310

/22/03

10/23

/0310

/24/03

10/27

/0310

/28/03

10/29

/0310

/30/03

10/31

/03#

of S

ched

uled

Cas

es

E6W Direct Nursing Hours per Patient Day

8.66

8.16

7.90

8.00

8.10

8.20

8.30

8.40

8.50

8.60

8.70

Prior to Vascular SmoothingAfter Vascular Smoothing

Average CT Surgery Unscheduled Cases Weekdays

Average Scheduled CT Surgery Cases by Weekday

Cardiac Scheduled Cases Histogram January & February Non-holiday Weekdays Only

0%

5%

10%

15%

20%

25%

30%

35%

Monday Tuesday Wednesday Thursday Friday

20042003

March Daily PCU Census - 2003 vs. 2004

0

2

4

6

8

10

12

1-Mar

2-Mar

3-Mar

4-Mar

5-Mar

6-Mar

7-Mar

8-Mar

9-Mar

10-M

ar11

-Mar

12-M

ar13

-Mar

14-M

ar15

-Mar

16-M

ar17

-Mar

18-M

ar19

-Mar

20-M

ar21

-Mar

22-M

ar23

-Mar

24-M

ar25

-Mar

26-M

ar27

-Mar

28-M

ar29

-Mar

30-M

ar31

-Mar

20032004

2003 range 10 – 1 = 9

2004 range 7 –2 = 555% reduction in variability

Operating Outside of theBlock at BMC

Separating the Flow of Elective Surgery from

Urgent/Emergent Surgery

Menino Pavilion compared to Newton PavilionVariable NP MP

# Rooms 13 8

# Cases Day 30-35 25-32

# Cases Year 8601 6608

Cancellation Rate 10% 20%

#Add Ons Per Day 1-2 5-12

#Weekend Cases 0-4 5-20

Unique Services Cardiac, Ophth Pediatrics, Trauma, Gastric Bypass, OB

Pre-change Problems with the Daily Schedule – Menino Pavilion

•Urgent/emergent bump elective cases•Overall 50% block utilization•Variable use of block (vacation,meetings)•Most cases booked 3-4 days out•33% of daily schedule is “add ons” •Variable release time between services•Cases can be lost waiting•People live in fear of losing their block

The Radical Changes

#1 Eliminated Block Booking

#2One Urgent Room Created

OR 5

Bumped Cases Before and After Separating “Flows”

After

April 04 – April 05

• 354 emergent cases (M – F) 7:00 AM to 3:30 PM

• 7 elective patients were delayed or cancelled

Before

April 03 – April 04

• 349 emergent cases (M – F) 7:00 AM to 3:30 PM

• 771 elective patients were delayed or cancelled

Norwood: Biggest Operational Dilemma

Daily ED Admits and Time from Decision to Departure

0100200300400500

1/1/20061/3 /20061/5 /20061/7 /20061/9 /2006

1/11/2006

1/13/2006

1/15/2006

1/17/2006

1/19/2006

1/21/2006

1/23/2006

1/25/2006

1/27/2006

1/29/2006

1/31/2006

2/2 /20062/4 /20062/6 /20062/8 /2006

2/10/2006

2/12/2006

2/14/2006

2/16/2006

2/18/2006

2/20/2006

2/22/2006

2/24/2006

2/26/2006

2/28/2006

Date

Min

utes

01020304050

Adm

issi

ons

Time in MinutesNumber of Admits

TimeRange = 176 – 418 minutes (3 hours – 6 hours)

Mean = 300 minutes (5 hours)Number of ED Admits

Range = 23 – 45Mean = 30

Goal = 120 minutes

What is the true constraint?Physician workup in the ED.

Find it and elevate it.Moved to the inpatient unit.

What is now the true constraint?Floor not ready.

Find it and elevate it.Create Transfer Time.

Some other constraints

• No transporter: included transport in synchronization and added transport capacity

• No nurse to staff an inpatient bed: stopped staffing to monthly historic mean; create prediction software based on historic natural variability and today’s census for tomorrow

1. ED MD evaluates new patient and decides that this is a medical patient (non-ICU) that needs to be

admitted.

11. Is a bed likely available for the patient?

3. ED MD completes admission order and

submits to ED secretary.

4. ED secretary enters order into Meditech,

which generates print out in Admitting office.

8. ED MD signs out patient to Lead Hospitalist (LH).

6. ED secretary informs ED charge RN of admission.

7. ED charge RN denotes admission

on white board.

5. ED secretary updates admission log.

ED Medical Admissions Process: IN THE EMERGENCY DEPARTMENT...

2. ED MD informs ED primary RN of

admission (probably when informing pt.).

9. ED MD informs charge RN of LH contact (during

board run).

12. Bed availability summary

sent by BPC

NO

YES

13. ED charge RN instructs primary RN to tape voicemail report.

14. ED charge RN fields call from BPC confirming

bed and transfer time.

15. ED charge RN puts bed and transfer time on board.

16. ED charge RN informs primary RN of bed and transfer time.

17. Primary RN tapes voicemail report (if not already done).

20. ED secretary fields call from BPC with bed and

transfer time. **If bed and transfer time are not

written on the white board, ED secretary makes sure ED charge RN is aware of

bed and transfer time.

21. ED secretary enters bed and transfer time on admission log.

19. Primary RN prepares patient for transfer, copies chart, and places copy

of chart in order bin.

10. ED charge RN denotes LH contact on white board

(during board run).

18. Primary RN calls receiving floor to deliver the voice-mailbox number.

22. Transport arrives to move the patient and collects necessary paperwork.

23. Transport moves patient out of ED.

E

G

B

24. ED secretary uses copy of chart to depart the patient in Meditech,

and to transfer the patient from U10 to the receiving unit.

A**The secretary should use the

“special comments” field to denote information that is necessary for bed placement (e.g.: “1:1”, “not

suitable for U31”, etc.).

Page 1

I

H

ED Medical Admissions Process: THE BED PLACEMENT COORDINATOR

(25) When Admit Order prints out, BPC:a) enters name, etc. into BPC logb) admits pt. in Meditechc) prints packet to ED printerd) sends text to Adm. Mgr. and Admissions RN (11a-11p) which includes name, age, Dx, bed type, potential unit (e.g. ?U31), precautions or 1:1 if known

If 120 minutes elapse from the time of admission order without contact

from the LH, BPC sends a text page to the LH to ask about the patient.

(26) BPC receives text from Lead Hospitalist, which includes: pt. name, diagnosis, bed type, ESI, and acceptance time.BPC enters time of page and ESI into log.

(27) If there are any discrepancies or questions about an admission, BPC contacts the administrative manager. Otherwise, BPC chooses a bed for

the patient.

(28) BPC calls charge RN (or secretary if charge not

available) on receiving unit to inform of admission (including

Dx and ESI) and to confirm bed assignment and transfer time.

(29) BPC calls ED charge RN (or sec. if charge not avail.) to confirm bed assignment and

transfer time.

(30) Once bed assignment and transfer time are confirmed, BPC

sends a text page to Administrative Manager,

Admissions RN (11a-11p), ED Admitting, and Lead Hospitalist

with the following info:a) pt. name

b) bed assignmentc) transfer time

Either the receiving unit or the ED may request a change in the transfer time.

If this happens, BPC contacts the opposite unit (e.g. if ED calls, then

BPC calls the receiving unit) to confirm new transfer time. BPC notes the

changed transfer time in the log and then resends a text page to

Administrative Manager and Lead Hospitalist beginning with “Change:”

and then the pt. name, bed, and transfer time.

(31) BPC calls ED secretary to deliver bed assignment

and transfer time.

If the bed type is changed (whether by ED order or LH contact), BPC alerts the Administrative Manager

by text page.

Our target is to set the transfer time within 30 minutes of the call to the

floor. When contacting the receiving unit, the BPC should state, “We’d like

to send this patient up in 30 minutes. Is there any chance we

could send the patient sooner?” The floor reserves the right to request a

transfer time greater than 30 minutes, but must inform the BPC of the reason.

A

C

D

E

G

Page 2

(32) BPC checks to ensure that patient moves upstairs on time. If patient is still in ED after transfer

time, BPC contacts ED charge RN.

F

ED Medical Admissions Process: HOSPITALISTS AND INPATIENT UNIT

(33) Lead Hospitalist (LH) receives sign-out and determines when a hospitalist can see the patient.

(34) LH communicates the following to Bed Placement Coordinator: pt.

name, diagnosis, bed type, ESI, and acceptance time.

(35) Assigned hospitalist begins work-up (regardless of patient location) as soon as possible.

(36) LH receives bed assignment and transfer time from BPC.

(37) Assigned hospitalist proceeds to floor as soon as

possible.

(42) Patient arrives on floor.

B

D

C

F

I

Page 3

(43) If patient does not have admission orders, unit secretary

texts Lead Hospitalist that patient has arrived.

H

(38) Charge RN or unit secretary on receiving unit fields call from BPC and

establishes bed and transfer time.

(39) Charge RN informs receiving RN of admission.

(40) Unit secretary (??) fields call regarding report

and informs receiving RN of mailbox number.

(41) Receiving RN retrieves voicemail report and calls

ED primary RN for clarification if necessary.

ED Time from Decision to Departure and Total ED LOS (admits)

0100200300400500600700

1/1/2 0061/8/2 006

1/15/ 20061/22/ 20061/29/ 2006

2/5/2 0062/12/ 20062/19/ 20062/26/ 2006

3/5/2 0063/12/ 20063/19/ 20063/26/ 2006

4/2/2 0064/9/2 006

4/16/ 2006

Date

Min

utes

Redesign Go live

SERVICE

Reduce the average time from ED admission decision to departure to inpatient unit to 120 minutes calculated monthly.

Av g. T im e from Decis ion-to-Adm it to ED Departure

0

50

100

150

200

250

300

350

6/19/2

0057/1

9/2005

8/19/2

0059/1

9/2005

10/19

/2005

11/19

/2005

12/19

/2005

1/19/2

0062/1

9/2006

3/19/2

0064/1

9/2006

5/19/2

0066/1

9/2006

7/19/20

068/1

9/2006

9/19/2

00610

/19/200

611

/19/200

6

Week B eg inn ing ...

Min

utes

D TA -D EP

Question Mean Score: Speed of Admission

Key change concepts of the Design

• Do tasks in parallel: move the patient to the floor while the workup continues

• Synchronize: assign a transfer to floor time (creates pull) after communication with charge nurses and hospitalist

• Central command: all beds are assigned by the nursing supervisor/bed facilitator

• Direct Communication: ED physician hands-off to Hospitalist

• Predict Demand: Use data on natural variability to get ready for staffing changes

Summary• There is much artificial variability in healthcare. We can no

longer afford this waste.• We must redesign our systems to maximize flow which will

make our patients safer, improve volume, staff and patient satisfaction and reduce the waste.

• Separating the flow of urgent surgery from scheduled surgery reduces waste and rework.

• No-Block scheduling is a good way to help the surgeons, patients, and staff.

• All hospitals should map inpatient flow and test changes to improve it.

References

• The Goal; by Eliyahu Goldratt• Leading Change; by John P. Kotter• The Improvement Guide; by Langley et al• http://management.bu.edu/research/hcmrc/mvp/index.asp• Rathlev NK, Chessare J, Olshaker J, Obendorfer D, Mehta

SD, Rothenhaus T, Crespo SG, Magauran B, Davidson K, Shemin R, Lewis K, Becker JM, Fisher L, Guy L, Cooper A, Litvak E. Time Series Analysis of Daily Emergency Department Length of Stay. Ann Emerg Med 2007;49:265-271