1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew...

45
1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie University PO Box #1000 Halifax, NS B3J 2X4 CANADA

Transcript of 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew...

Page 1: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

1

Waitlist Management in Nova Scotia:Policy and Practice

John T. Blake, Peter VanBerkel, Matthew Campbell

Department of Industrial EngineeringDalhousie University

PO Box #1000 Halifax, NS B3J 2X4CANADA

Page 2: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

2

Nova ScotiaNova Scotia

• Canada’s 2nd smallest province (55,000 km2)

• About 950,000 people• Principle cities: Halifax

(310k) and Sydney (110k)• Principle industries:

Government services, finance, retail, manufacturing, forestry

• About 5800 km from Wroclaw

• About 60 south of Wroclaw

Page 3: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

3

How’s The Weather?

Wroclaw - Winter 2005

www.pbase.com/tygrys50/your_favorite

Halifax - Winter 2005

Page 4: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

4

Healthcare in Nova Scotia• 43 hospitals

– 1 adult tertiary (on two sites)– 1 paediatric– 9 regional/mid-size centres

• 2000 physicians– 700 medical residents and

interns– 60% on fee for service

• Budget of $2.6 billion ($Can)

• Provides tertiary services to patients from 3 Atlantic provinces.

Page 5: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

5

Problem Statement

• Perception amongst docs & patients that wait times for elective procedures are long.

• There has been a lot of “buzz” around wait times in Canada in general and in Nova Scotia.

• There have been some recent rumblings in the press and policy areas about wait times.

• More recently, there has been a supreme court ruling that may (or may not) change the nature of the CHA.

Page 6: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

6

Some Anecdotal Evidenceon Wait Lists

CBC (Online Edition): Jan 23, 2003

www.npdcaucus.ns.ca

Page 7: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

7

The Latest Wrinkle…

• Recent supreme court ruling ties wait time to charter freedom

• Appears to allow the introduction of private insurance or services

• Could signal a major shift in Canadian health care

• Most provinces have adopted wait time or access management commissions

Canadian Medical Association June 9/05

Page 8: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

8

Finally Some Action?Federal Advisor on Wait Times

• Under a funding “bump” in 2004, the provinces and the feds agreed to some interesting mechanisms to manage waits.

• The central idea is that research is required to understand the root causes of waits and cost-effective methods of resolution

• Areas for review include:1. Development of benchmarks for access, suitable to the Canadian context

2. Develop criteria for appropriateness

3. Identify the nature and causes of wait times, including physical capacity, process flow efficiency, spatial-geography issues, and barriers to care

4. Use operational research (!) to improve productivity and quality

5. Examine the impact of organizational design, policies, and incentives on wait times

6. Look at the impact of the media on the perception of wait times

Page 9: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

9

So what do we know about waits?

With rare exceptions, waiting lists in Canada, as in most countries are non-standardized, capriciously organized, poorly monitored, and (according to most informed observers) in grave need of retooling

McDonald, Shortt, Sanmartin, Barer, Lewis and Sheps (1998)

As such, most of those currently in use are at best misleading sources of data on access to care, and at worst instruments of misinformation, propaganda, and general mischief

Page 10: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

10

Rationing by Wait List:Are Waits Always Bad?

Positives• Equity: Time is more equally

distributed than cash• Broadly seen as equitable

within a societal context• Discourages consumption

where social costs outweigh social benefits

Negatives• Masks mismatches between

supply and demand• Those who receive may not

be most deserving• Efforts to stratify by need

subject to capriciousness• System can gamed; wealthy

more able to access care or bypass system

Wait times represent a non-price form of rationing healthcare

Page 11: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

11

Summary of Canadian Wait List Initiatives

• Medical community views “wait list” initiatives as:1.Registry of patients waiting for surgery2.Prioritization scheme for ranking patients3.Management tool (i.e. data base)

• Since Canada lacks IT infrastructure to collect objective data, self-reported survey data is commonly used

• Most effort (and cash) has been expended on prioritization or triage tools

• The understanding of the need for objective data and an underlying conceptual model for wait times is just developing

Page 12: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

12

Registry Methods: BC

0

2

4

6

8

10

12

14

16

Weeks

Dent

Opht

Gyna

Orth Urol

Card

Cance

r

Median Wait Times

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

20,000

Dent

Opht

Gyna

Orth Urol

Card

Cance

r

Patients on List

A 2005 audit found ~8,000 of 68,000 patients either redundant, double counted, dead, or no longer in need of surgery

Page 13: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

13

More Policy QuestionsCan We Believe What we See?

• Are wait lists inherently biased?– Physician induced demand: Docs may have an incentive to

over prescribe, particularly if funded on FFS– Backwards bend supply curve: Longer wait lists seen as a

sign of prowess– Your money or your life: Inflating wait lists may be a way to

secure additional funding or hospital resources– Double counting: Patients may be double counted, dead, or

no longer in need of services– Gate keeper: Ultimately physicians, not patients, make

decision about who is/is not on wait list

Page 14: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

14

Which Numbers are Correct?

BC Median Wait Time

0

10

20

30

40

50

60

THRTKR

Corne

aO

rthPla

stENT

Card

Catar

act

Vasc

Oph

t

Gyn

aNeu

rG

en Urol

Rad O

nc

We

eks

BC MoH

Fraser Inst

Page 15: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

15

• Even if we spend more, there is an inconsistent relationship between spending and performance metrics

• Latent Demand (or “A built bed is a filled bed”): – As we provide more resources, barriers to entry are lowered– Wait time, typically, decreases– This allows the procedure to be more widely prescribed– Thresholds for appropriateness drop– Gradually the system returns to its congested state

• A number of Fraser Institute reports suggest that wait time is not correlated with increased spending– Increased institutional spending actually increased wait– Wait was seen to decrease with increased physician spending

• Similar findings are reported in the UK

The Dilemma: Getting bang for your Buck

Page 16: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

16

International Experience: UK

• The UK has perennially had issues with wait time

• Over the past five years, however, wait for elective procedures has dropped

• Reductions appear to be in response to a fiat on maximum waiting times

• There is some indication that long waits have decreased, but average waits are largely unchangedSource: King’s Fund Trust

Page 17: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

17

The Operational Research Perspective: This is an easy problem! Isn’t it?!?

• People have been studying line ups for about 100 years.• Much of the original work was done in relation to telephone

switches.• With some assumptions we can fully define the operation of a

queue with three or four pieces of data:

Arrival Rateλ (Customers/Hour)

Server Rateμ (Customers/Hour)

Number of Servers (s)

Queue Size(usually infinite)

Queue Discipline(usually FIFO)

Page 18: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

18

Some Basic Results from OR

Queue Length vs Traffic Intensity

0

2

4

6

8

10

12

14

16

18

20

0 0.2 0.4 0.6 0.8 1

Lambda/Mu

Pat

ien

ts

Queue Length vs Number of Servers

0

0.05

0.1

0.15

0.2

0.25

0.3

0 5 10 15 20

Servers

Qu

eu

e L

en

gth

Queue Length vs Traffic Intensity and Service Variance

Sigma = 0.5Sigma = 1

Sigma = 2

0

5

10

1520

25

30

35

40

0.7 0.75 0.8 0.85 0.9 0.95 1

Traffic Intensity

Qu

eu

e L

en

gth

Queue Length vs Traffic Intensity and Queue Discipline

FIFOLIFO

LPT

SPT0

5

10

15

20

25

30

35

0.70 0.75 0.80 0.85 0.90 0.95 1.00

Traffic Intensity

Qu

eu

e L

en

gth

Page 19: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

19

The $64k Question:Why Isn’t OR in Greater Use in Healthcare?

• Timing/Project Cycle: – Simulations typically take a long time to build and validate– Issues tend to be “front burner” for institutions

• Cost– Simulation requires specialist knowledge & software

• Data Availability:– It isn’t– IT systems are designed for clinical and administrative

purposes; patient flow hasn’t been a design issue– In the Canadian context, process management is seen as

administrative overhead.

Page 20: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

20

Why Ask Us?

• Like most places, Nova Scotia currently lacks complete data to make an accurate determination of wait

• It does have integrated billing and discharge data and is relatively compact

• We’ve been asked to look at efficiency aspects of access

• We do have some experience in orthopaedics

• I’ll talk about some of our work in DI, surgery, and a provincial model

6682

98 0%10%

20%

0

0.2

0.4

0.6

0.8

1

Wait Time (Years)

Beds Increase in OR Time

Ortho Wait Times as a Function of Beds and OR Time

CDHA Ortho (’04)

Page 21: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

21

Nova Scotia Access Plan

1. Invest in efficiency improvements first

2. Streamline and simplify the process

• Develop patient centred care• Reduce no-shows

3. Make DHA’s accountable for improving timely access to care

4. Adopt evidence based decision-making

5. Measure clinical and administrative outcomes

6. Manage access to services better• Standard triage tools• Centralized wait lists

7. Communicate access data with the general public

8. Invest in IT strategies9. Increase capacity only

when efficiency gains have been exploited.

10.Develop integrated health human resources plans

Page 22: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

22

DI Issues

• Some of our contributions have been logistical in nature– Implemented better data

collection methods

– Adapted QC tools for analysis

• Some of our contributions are in the area of models– DEA analysis of providers

– A general rant on the 3rd available appointment slot(ongoing)

XmR Chart for MRIDHA XX Un-Named Regional Hospital

0

10

20

30

40

50

60

70

Dec-04

Jan-

05

Feb-0

5

Mar

-05

Apr-0

5

May

-05

Jun-

05

Jul-0

5

Aug-0

5

Sep-0

5

Oct-05

Nov-05

Dec-05

Jan-

06

Feb-0

6

Mar

-06

Wai

t

• The province now uses this tool to identify institutions that are out of control

• Results are reported back to managers and institution CEOs

• Implemented quarterly meetings with DI managers to review results

Page 23: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

23

Diagnostic Imaging DEA Study

• There are 36 providers in NS – 2 tertiary; 9 regional; 27 rural• We run separate analysis for each band• Potential inputs include

– Budget √– Staff √– Numbers & types of machines

• Potential outputs include– Number and types of tests √– Workload units √

• We produce both efficiency scores & comparator institutions• We think we are the first people to apply DEA across DI

departments within a province• Implementation of the CCRD-I model with constant returns

Page 24: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

24

DEA Example: RuralEfficiency

Number Include DMU Name Score Budget Techs Radiography US Workload Units1 1 Hospital 01 0.338 0.22 2.51 0.559 0.000 0.5672 1 Hospital 02 0.816 0.28 3.54 1.527 0.000 1.8573 0 Hospital 03 0.00 0.00 0.000 0.000 0.0004 1 Hospital 04 0.879 0.31 3.31 0.814 1.481 1.4855 1 Hospital 05 0.441 0.17 2.52 0.567 0.000 0.5636 0 Hospital 06 0.00 0.00 0.000 0.000 0.0007 1 Hospital 07 0.414 0.13 1.70 0.403 0.000 0.3988 1 Hospital 08 1.000 0.03 0.16 0.204 0.000 0.2249 1 Hospital 09 1.000 0.43 4.17 1.337 2.550 2.28910 0 Hospital 10 0.00 0.00 0.000 0.000 0.00011 1 Hospital 11 0.749 0.08 0.63 0.463 0.000 0.48112 0 Hospital 12 0.00 0.00 0.000 0.000 0.00013 1 Hospital 13 0.796 0.10 1.19 0.277 0.000 0.66114 1 Hospital 14 0.564 0.15 2.25 0.617 0.000 0.68615 0 Hospital 15 0.00 0.00 0.000 0.000 0.00016 1 Hospital 16 0.365 0.11 1.41 0.290 0.000 0.26117 1 Hospital 17 0.195 0.05 0.74 0.074 0.000 0.06918 0 Hospital 18 0.00 0.00 0.000 0.000 0.00019 1 Hospital 19 0.146 0.07 0.95 0.077 0.000 0.07620 1 Hospital 20 0.204 0.08 1.19 0.016 0.000 0.14121 0 Hospital 21 0.00 0.00 0.000 0.000 0.00022 1 Hospital 22 0.138 0.06 0.86 0.057 0.000 0.05423 1 Hospital 23 0.274 0.28 2.17 0.568 0.000 0.60524 1 Hospital 24 0.524 0.08 0.92 0.264 0.000 0.35225 0 Hospital 25 0.00 0.00 0.000 0.000 0.00026 1 Hospital 26 0.758 0.44 4.10 1.042 2.000 1.77127 1 Hospital 27 0.388 0.12 1.20 0.340 0.000 0.39328 1 Hospital 28 0.541 0.11 1.23 0.443 0.000 0.46729 1 Hospital 29 0.570 1.27 9.85 5.017 0.000 5.97930 0 Hospital 30 0.00 0.00 0.000 0.000 0.00031 1 Hospital 31 0.373 0.17 0.39 0.440 0.000 0.51532 1 Hospital 32 0.557 0.72 6.77 2.354 0.000 3.29533 1 Hospital 33 0.391 0.07 0.75 0.219 0.000 0.19834 0 Hospital 34 0.00 0.00 0.000 0.000 0.00035 1 Hospital 35 0.478 0.13 0.52 0.461 0.000 0.43636 0 Hospital 36 0.00 0.00 0.000 0.000 0.000

1 1 1 1 1Total Efficient DMUs 2

Data Inputs Outputs

Include Element?

Units and institutions coded

Page 25: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

25

DEA Results

• Tertiary – not enough sites for meaningful analysis• Regional – Identified a single institution as benchmark• Rural – A bit more difficult

– We did identify two institutions in one DHA, managed by the same team that shows up as efficient on most subsets of inputs and outputs

• Implementation– Oddly, DI managers are a bit reluctant to talk efficiency – especially

in front of the province– Would really just like more money (If I’m efficient does that mean I

won’t get a new MRI machine?)– However, we are in the process of cleaning up data issues and

establishing a benchmark procedure

Page 26: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

26

Surgical Issues

• Our contributions are largely in the area of models– Evaluation of guaranteed

waits for surgery to meet federal benchmarks

– An analysis of general surgery at the QEII (the largest hospital in the province)

– The development of a general acute care model for all hospitals in the province

DHA 4 Dispose LTC DHA 4

DHA 4

DHA 4

RegionalColchester

DHA 4

MemorialLillian Fraser

DHA 4

B ayview Memorial

InptsDispose DHA 4

0

0 . 0 6 0 . 00. 0

1. 0

0

0

0 . 0 6 0 . 00. 0

1. 0

0

0 . 0 6 0 . 00. 0

1. 0

0

0 . 0 6 0 . 00. 0

1. 0

0

Page 27: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

27

Guaranteed Waits

• A number of ideas have been floated to deal with the impact of the supreme court ruling on waits.

• One of the more popular idea is the “guaranteed wait”• Patients would be separated into three broad bands.• After a fixed amount of time patients would be

“upgraded” into the next higher band• Some plans call for an automatic jump to the top band

• Is this likely to be an effective policy?

Page 28: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

28

Guaranteed WaitsAvg Lost Utility (r = 0.3)

0

20

40

60

80

100

120

0.7 0.75 0.8 0.85 0.9 0.95

Rho

QA

LD

With Guarantee

Without

Simple M/M/s model with an assumed exponential decay for utility

We conclude that guaranteed waits may result in greater lost utility – should address capacity issues up front.

Guaranteed waits are particularly dangerous if ρ > 1

Page 29: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

29

General Surgery Wait Time

• A discrete event simulation model using

– Modular design elements– Self building concepts– Excel interface to model

elements

• Data elements derived from three local sources

– Subject to a substantial level of cleaning and organizing

• Validated against a two year data sample for:

– Occupancy rate– Expected wait time– Patient LOS

Expected Wait Time for Elective Surgery

0

20

40

60

80

100

120

140

160

Date

Wai

t (da

ys)

Known Wait Time Modeled Wait TimeLinear (Modeled Wait Time)

Page 30: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

30

General Surgery: Bottleneck Analysis

Less 15%

Current

Plus 15%

(Current) 4

145

50210

215

220

225

230

235

240

Thruput (pnts/month)

OR Time

VG Beds

Through Put (Pnts/Month)

• Two-way design to look at factors limiting patient flow

• Analysis shows that beds, rather than OR time is the limiting factor

• However, system is sensitive to reductions in OR time

• Analysis showed a number of process issues – turn around being the most obvious

Page 31: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

31

Provincial Flow Model: ObjectivesConventional wisdom claims that the ED is backed up because inpatient beds are used by people who should be in nursing homes.

There has been a renewed call in the province for greater LTC beds

We have been asked to determine best bang for buck in terms of resources. Should we invest in:

1. Long term care beds

2. Acute care beds

3. Emergency services

4. All three

And if so, in what proportion?

Is a system wide fix required, or do local conditions dictate local approaches

Page 32: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

32

Project Methodology

• We are developing a simulation model of the entire province

– This is probably unique in Canada• The model will runs on a DAD

abstract for 2004/05• We have detailed models of acute

care, with simple extensions for LTC and (eventually) ED

• A phased approach to model building, testing, and development will be necessary

• We have developed a Phase 1 model (right) and are now working on extensions.

• The model is based on ARENA templates to reduce coding and repetition

DHA 4 Dispose LTC DHA 4

DHA 4

DHA 4

RegionalColchester

DHA 4

MemorialLillian Fraser

DHA 4

B ayview Memorial

InptsDispose DHA 4

0

0 . 0 6 0 . 00. 0

1 . 0

0

0

0 . 0 6 0 . 00. 0

1. 0

0

0 . 0 6 0 . 00. 0

1. 0

0

0 . 0 6 0 . 00. 0

1. 0

0

Page 33: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

33

Provincial Model: Data Items

• Of the 98,000 discharges only 2253 had any ALC days

• ~88,800 ALC days in the province out of 820,000 inpatient days

• ~138,000 days consumed by patients who ultimately end up with an ALC day

Patient Volume

0

5000

10000

15000

20000

25000

30000

Qu

e

e

n

El

i

z

a

b

e

t

h

II

He

al

t

h

Sc

i

e

n

c

e

Ce

n

tr

e

IW

K

H

ea

l

t

h

C

e

n

t

re

Ca

p

e

Br

e

t

o

n

He

a

l

t

hC

ar

e

C

o

m

pl

e

x

Va

l

l

ey

Re

g

i

o

n

a

l

H

os

pi

t

a

l

Ab

e

r

de

e

n

Ho

s

p

i

t

a

l

Co

l

c

he

s

t

e

r

R

e

gi

o

na

l

Ho

s

p

i

t

a

l

Da

r

t

mo

u

t

h

Ge

n

e

r

a

lH

os

p

i

t

al

Ya

r

m

ou

t

h

Re

g

i

o

n

a

lH

os

p

i

ta

l

.

St

Ma

r

t

h

a

'

s

R

e

g

io

n

a

l

Ho

sp

i

t

a

l

So

u

t

hS

ho

r

e

Re

g

i

o

n

a

l

Cu

m

b

er

l

a

n

d

Re

g

i

o

na

lH

e

a

lt

h

Ca

r

e

Ce

n

tr

e

In

v

e

rn

e

s

s

Co

n

s

o

l

i

d

a

t

ed

Ho

s

p

i

ta

l

So

l

d

ie

r

s

'

M

e

m

or

i

a

l

Ho

sp

it

a

l

Ha

n

t

s

Co

m

m

u

n

i

t

y

H

os

pi

t

a

l

Qu

e

e

ns

Ge

n

e

r

a

l

H

o

s

p

it

a

l

No

v

a

Sc

o

t

i

a

H

o

s

p

i

ta

l

Ro

s

e

wa

y

Ho

s

p

i

t

a

l

Vi

c

t

or

i

a

Co

u

n

t

y

M

em

or

i

a

l

H

o

s

pi

t

a

l

Li

l

l

ia

n

Fr

a

s

e

r

M

e

mo

ri

a

l

H

o

s

pi

t

a

l

Di

g

b

y

Ge

n

e

r

a

l

Ho

s

pi

ta

l

Fi

s

h

er

m

e

n

M

e

m

o

r

i

al

An

n

a

po

l

i

s

Co

m

m

u

n

it

yH

e

a

lt

h

Ce

n

t

r

e

Bu

c

h

an

a

n

Me

m

o

r

i

a

l

Ho

s

p

i

t

a

l

-

St

r

a

it

Ri

c

h

m

o

n

d

Ho

sp

i

t

a

l

Sa

c

r

ed

H

e

a

r

t

H

o

s

p

it

a

l

Ea

s

t

er

n

Sh

o

re

M

e

m

or

ia

l

H

o

sp

i

t

a

l

Gu

y

s

bo

r

o

u

g

h

Me

m

o

ri

al

Ho

s

p

i

t

al

No

r

t

h

Cu

m

b

e

r

l

a

n

d

Me

mo

r

i

a

l

H

o

sp

i

t

a

l

Tw

i

n

Oa

ks

M

e

m

o

ri

a

l

Ho

s

pi

ta

l

Mu

s

q

u

o

d

ob

o

i

t

Va

l

le

y

Me

m

o

r

i

a

l

Ho

s

p

i

t

a

l

.

St

Ma

r

y

'

s

M

e

m

o

r

i

al

Ho

s

p

i

t

a

l

Ea

s

t

er

n

Me

m

o

r

i

a

l

Ho

sp

i

t

a

l

Ba

y

v

ie

w

Me

m

o

r

i

a

l

He

al

t

h

C

e

n

t

re

So

u

t

hC

u

m

b

e

r

l

an

d

Co

m

m

un

it

y

Ca

r

e

C

e

n

tr

e

Pat

ien

ts

All Patients

ALC Patients

Institution names obscured

Page 34: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

34

Is ALC the Only Factor?

0

5000

10000

15000

20000

25000

30000

35000

40000

SC

HIZ

OP

HR

EN

IA A

ND

OT

HE

R P

SY

CH

OT

IC

OT

HE

R F

AC

TO

RS

CA

US

ING

HE

AR

T F

AIL

UR

E

SP

EC

IFIC

CE

RE

BR

OV

AS

CU

LAR

ES

OP

HA

GIT

IS,

GA

ST

RO

EN

TE

RIT

IS

SIM

PLE

PN

EU

MO

NIA

AN

D P

LEU

RIS

Y

CH

RO

NIC

BR

ON

CH

ITIS

OT

HE

R S

PE

CIF

IED

AF

TE

RC

AR

E

MA

JOR

INT

ES

TIN

AL

AN

D R

EC

TA

L

DE

PR

ES

SIV

E M

OO

DD

ISO

RD

ER

S W

ITH

OU

T

CH

RO

NIC

OB

ST

RU

CT

IVE

DE

ME

NT

IA W

ITH

OR

WIT

HO

UT

DE

LIR

IUM

NE

ON

AT

ES

WE

IGH

T >

2500

GM

(N

OR

MA

L

GA

ST

RO

ST

OM

Y A

ND

CO

LOS

TO

MY

OT

HE

R A

DM

ISS

ION

SW

ITH

SU

RG

ER

Y

VA

GIN

AL

DE

LIV

ER

Y

DE

ME

NT

IA W

ITH

OR

WIT

HO

UT

DE

LIR

IUM

RE

SP

IRA

TO

RY

NE

OP

LAS

MS

AM

I WIT

HO

UT

CA

RD

IAC

CA

TH

To

tal D

ay

s Conservable

ALC

Expected LOS

This chart suggest that while ALC bed days are an issue, so too are “conservable” days

Page 35: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

35

Nevertheless LTC Admissions are tight

LTC Transfers per Day

0.0000

0.2000

0.4000

0.6000

0.8000

1.0000

1.2000

DHA1

DHA2

DHA3

DHA4

DHA5

DHA6

DHA7

DHA8

DHA9

Pts

/Day Community

Hospital

Overall, only 3 patients per day can be transferred from Acute Care to LTC in NS

Page 36: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

36

Some Interesting Notes • This is, to the best of my knowledge, the only system wide in existence in Canada or anywhere.• We are modelling at a high level, but the framework is very flexible and easily extended• The model should be seen as an evolutionary entity – we are starting simple and building up

confidence and capability in the model and its results

Page 37: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

37

Phase 1: Model

• A single bed complement for each acute care facility

• A single patient type with a common LOS distribution

• All ALC patients transfer to LTC facilities

• Assume a single LTC facility for each DHA

• LTC facilities take admissions from community and acute care institutions

• All model widgets are “self-contained” instances of a generalized process

DHA 1 DHA 2

DHA 3 DHA 4

1Dis pos e LTC DHA

2Dis pos e LTC DHA

3Dis pos e LTC DHA

4Dis pos e LTC DHA

DHA 1

D H A 1

M em or ialFisher m ans

I np t sDis pos e DHA 1

D H A 1

Q ueens G ener al

D H A 1

RegionalSout h Shor e

D H A 2

Digby G ener alI np t s

Dis pos e DHA 2

D H A 2

Ros eway Hospit al

D H A 2

RegionalYar m out h

DHA 2

D H A 3

Cent r eCom m unit y Healt h

Annapolis

D H A 3

Soldier s M em or ial

D H A 3

Valley Regional

I np t sDis pos e DHA 3

DHA 3 DHA 4

D H A 4

RegionalColc hes t er

D H A 4

M em or ialLillian Fr aser

D H A 4

Bay v iew M em or ial

I np t sDis pos e DHA 4

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0 0

0 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0 0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

0 . 0 6 0 . 0

0 . 0

1 . 0

0

Page 38: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

38

Phase 1: ValidationHow do we test the model is working?

300250200150100500

Median

Mean

86420

Anderson-Darling Normality Test

Variance 420.114Skewness 5.5617Kurtosis 43.4584N 4187

Minimum 0.000

A-Squared

1st Quartile 0.097Median 0.9823rd Quartile 5.814Maximum 322.244

95% Confidence Interval for Mean

7.230

822.52

8.472

95% Confidence Interval for Median

0.848 1.086

95% Confidence Interval for StDev

20.067 20.945

P-Value < 0.005

Mean 7.851StDev 20.497

95% Confidence I ntervals

Summary for QEII LOS

• We test the averages for:- Arrival rates

- Inpatient length of stay

- Transfers to LTC

using standard statistical tests (t-test) and compare model results against samples from DAD

• We also test variance (σ2) using standard statistical techniques (χ2

test)

• We have no problem in reproducing admission numbers and appropriate lengths of stay.

Page 39: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

39

Now for the bad news…

• Having completed a 1st model, we know our admissions are correct

• We know that the LOS is correct• However, our bed utilization

numbers are too low– In almost all instances, our

model does not show a bottleneck

• Reasons could include– Home care excluded– OR time excluded– Inadequate patient

categorization– Fluctuations in bed availability

over the year– Transfers between institutions

0

0.2

0.4

0.6

0.8

1

1.2

AB DA DDDDDDDDDDDDD D ABAAAAAAAAAAAAA ADDDDDDDDDDDDD D D AD BI D BBBBBBBBBBBBB B BD L BB A

AD

BBBBBBBBBBBBBB BD

BD

Page 40: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

40

Patient disposition is also an issue

0

20000

40000

60000

80000

100000

120000

Nursinghome

Home care Acute Ambulatory Rehab Emerg Forensics Out of NS Out Pt Chronic Unknow n Daysurgery

Patient Days by Transfer

ALC

Conservable

PLEX

• Interestingly, 26% of ALC bed days are consumed by patients who ultimately go home.

Page 41: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

41

Issue: Appropriate Patient Types

MED_SURG

TOTA

L_STA

Y

unknownsurgicalpregnancyneonatemental hmedical

35

30

25

20

15

10

5

0

Interval Plot of TOTAL_STAY vs MED_SURG95% CI for the Mean

• There are statistically significant differences between med/surg and all other admission types

• Mental health, in particular, has a very long LOS

• We’ve decided to eliminate neonates – triggered by maternal admissions

Page 42: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

42

Issue: Admit Category(All significant except Mental Health)

ADMIT_CAT

TOTA

L_STA

Y

UL

13

12

11

10

9

8

7

6

5

4

95% CI for the MeanInterval Plot of TOTAL_STAY vs ADMIT_CAT (Surgical)

ADMIT_CAT

TOTA

L_STA

Y

UL

5.5

5.0

4.5

4.0

3.5

3.0

95% CI for the MeanInterval Plot of TOTAL_STAY vs ADMIT_CAT (Pregnancy)

ADMIT_CAT

TOTA

L_STA

Y

UL

50

40

30

20

10

95% CI for the MeanInterval Plot of TOTAL_STAY vs ADMIT_CAT (Mental Health)

ADMIT_CAT

TOTA

L_STA

Y

USNL

10.0

7.5

5.0

2.5

0.0

-2.5

-5.0

95% CI for the MeanInterval Plot of TOTAL_STAY vs ADMIT_CAT (Medical)

Page 43: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

43

Issue: Transfers

G

H

B

C

D

E

F

Other

RehabNursingHome

Out ofProvince

Homecare

A

Page 44: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

44

Phase II

• Home Care– We have added extensions to simulate

home care– We are collecting and validating capacity

and length of stay data

• Acute Care– We are implementing the model on a DHA

by DHA basis• This allows us to vet assumptions and

include local conditions

– We have added capacity for different patient types and admission categories

– We are now working to establish appropriate bed numbers and include surgical process capacity over time

• Long Term Care– One of our more difficult jobs at present is

validating LOS assumptions

Page 45: 1 Waitlist Management in Nova Scotia: Policy and Practice John T. Blake, Peter VanBerkel, Matthew Campbell Department of Industrial Engineering Dalhousie.

45

Future Plans

• Validate acute care modules• Expand model to include ED• Expand to model specific services (i.e. Ortho)• Include a specific widget to represent OR

time and master surgical schedule• “Package” simulation widgets• Develop a platform for local use of simulation

models• Target date: January 2007