Building a Decision Support Tool for
Delivery of Health Services for Hip and
Knee Osteoarthritis Patients
Taming of the Queue Workshop
March 28 2012
Deborah Marshall, PhD
Canada Research Chair, Health Systems and Services Research
University of Calgary
Alberta Bone and Joint Health Institute
Project Team
• Co-Investigators from U Toronto and U Calgary:
– Deborah Marshall (PI), Cy Frank, Tom Noseworthy, Sonia
Vanderby, Dianne Mosher, Mike Carter, Tom Rohleder, Paul
Rogers, Colleen Maxwell
• Partners:
–
–
–
–
Alberta Health Services
Alberta Bone Joint Strategic Clinical Network
Alberta Bone Joint Health Institute
Bone and Joint Canada
• Contributors: physicians, surgeons, health system
administrators, clinic managers, other clinicians
When is System View Useful in
Health Care?
• Many interrelated aspects of health care
– Population and societal factors are important and
“predictable” (aging population, disease morbidity)
• System Model for understanding vs. “answers”
– Model building identifies gaps in knowledge and data
– Process enhances understanding of relationships
– Modular approach to examine changes to system in context
of care delivery
– Use of models as “flight simulators” – analysis of health
policy options
Objective: Create a decision-support tool for strategic
service planning of care for OA patients Across care continuum and sustainable
 Tool to enable policy makers, service planners and
administrators, and clinicians to evaluate care
quality and system performance
 Balancing the tradeoffs between accessibility,
effectiveness and efficiency
 Inform choices about health system interventions
Goal is a sustainable plan for OA care
Why System Dynamics Model?
Health Care is a System and it is Dynamic
• Examines how to balance demand for H&K OA health services,
and supply and delivery of those services
• Considers resource constraints by changing the flow rate
through feedback loops (current demand and backlog)
• Conceptually represents the “big picture”
• Population-level care delivery
• Capacity, flow rates, utilization, wait times
• Captures changes projected over time
• Enable users to explore numerous scenarios and compare
results to inform decisions
Studying the Problem: Care Flows for Hip
and Knee Replacement
Demand for
Replacement
Inflow Rate
Stock
Patients waiting
for Replacement
Outflow Rate
Patients
Receiving
Replacement
4 Phase Development Plan
Phase 1
• ‘Proof of Concept’ Model
• Demand from OA arrivals (incidence & population projections)
• Flow rates affected by stage durations
• Patient routings based on historic proportions
Phase 2
• Phase I +
• Resource use at each stage
• Costs associated with resources
Phase 3
• Phase II altered
• Key resources as inputs with availability affecting flow rates
Phase 4
• Phase III +
• Factors and feedback loops affecting patient routes & flow rates
7
Phase I Development Process
• Literature review initial process diagram
• Expert feedback & input via 6 workshops
• Refine process diagram
• Identify resources throughout process
• Identify other factors affecting patient flow
Hip & Knee Working Group
• 3 workshops
• Focus: surgical
management process
Arthritis Working Group
• 3 workshops
• Focus: medical
management process
8
Overview of Care Process
9
System Model Diagram
Medical Management
Surgical Management
Self Treating
R - OA onset
Orthopaedics
S - Pts - Self
Treating
R - contact PCP
S - Pts awaiting PC
R - PC non
referred
S - Pts in OS
care - non
surgical
Rheumatology
R - to PC
S - Pts
staying in PC
R - PC
refer to
OS
S - Pts in PC
- referred
R - PC
refer to
RH
R - leave
- PC
R - Pts OS to PC
S - Pts
awaiting RH
S - Pts in RH
care
R - to
RH
R - ext MSK to PC
Primary Care
R - to
OS fro
m PC
S - Pts in OS
care - consult
R - RH to PC
S - Pts
awaiting OS
from RH
S - Pts
awaiting
MSK MD
R - OS
non surg
R - enter
extended
R - PC refer to
MSK MD
R - RH
refer to
OS
S - pts in PC
post
R - leave PC postspecialist
spec
S - Pts
awaiting OS
from PC
R - MSK
to PC
R - MSK
MD refer
R - pts to ext MSK to OS
S - Pts in
extended
MSK care
Rexternal
surg
R - extend
S - extendeded to surg S - external
wait
pts in OS ca
presurg care
re - consult
S - pts awaiting
surgery
R - OS to
Surg wait
S - Pts in OS care
S - Pts awaiting
- consult from
OS - from MSK
MSK
MD
R - to OS
from
MSK MD
R - pts await
2nd primary
- LT
R - OSm R - OSm - R - OSm
- wait
non surg extended
S - pts awaiting 2nd
primary >1 year
R - pts
await 2nd
primary - ST
Rexternal
to wait
R - to
OS
S - Pts in
MSK MD
care
R - to M
SK MD
R - depart PC non surg
S - Pts in PC non-surgical
S - pts awaiting
2nd primary <1 yr
R - pts to 2nd
primary ST
R - dep
art wait
S - pts to home
R - acute
to hc
S - Pts - home
with home
care
R - pts to
surgery
R - home yr1
R -depart hc
R - sub to hc
R - pts to 1st
surgery
R - other
to hc
R - depart sub
acute
S - Pts in acute care
R - Pts ext
MSK to
OS wait
Community Care
R - pts to 2nd
primary LT
R - acute
to home
R - depart
acute
R - acute to
sub acute
R - pts to
rev - ST
R - rev to surgery
R - other
to sub
S - pts to
rev in yr 1
S - LT revisio
n pts awaitin R - post
g surgery surg - to
rev ST
R - pts to rev longterm
S - Pts in sub
acute
R - sub
to other
S - Pts in other
facility
R - depart
other
Racute to
other
S - pts in LT
followup to rev
R - pts to
LT - rev
R - post
surgery no rev ST
S - pts in yr 1
post-surgery
R - pt
departures - ST
R - individ
uals to LT
R - pt
departure
s - LT
© DA Marshall, 2011
S - pts in long
term followup
Post-Surgical Follow-up
10
Overview of Data Flow
Confidential
© DA Marshall, 2011
11
Data Requirements and Sources
Model
Component
Information needed
Self directing
incidence rates
population projections
duration of time self treating
current population self treating
Primary Care
current population in primary care
duration of time in primary care
referral proportions
current population waiting for rheum & in rheum. care
duration of wait for consult & in rheum care
referral proportions
current populations waiting for screening & consult
current populations in screening & in surgeon care
Durations of wait times & time in care
referral proportions
current population waiting for surgery
duration of surgery wait & acute care LOS
discharge proportions
current community care populations
durations of time spent in community care
discharge proportions
current population post surgery
repair/revision rates
duration of time until revision & time remaining in system
Rheumatology
Orthopaedics
Acute care
Community
care
Rehabilitation
& Follow-up
Data source(s)
Alberta Health Population Registry
Ambulatory Care Classification System data
Canadian Primary Care Sentinel Surveillance Network
Discharge Abstract Database
Physician Payments
Statistics Canada Population projections
Survey of Living Chronic Disease in Canada
Physician Payments
Central Intake for Triage in Rheumatology
Physician Payments
Physician Payments
Hip and Knee Replacement Central Intake Clinic data
Operating Room Data
Discharge Abstract Database
Discharge Abstract Database
Home Care data ?
Discharge Abstract Database
Alberta Health Population Registry
Literature
12
Estimating OA Incidence and
Prevalence in Alberta
1.
Case Definition of OA
– Any diagnosis beginning code 715 or M15 to M19 (ICD9 & ICD10) in 4
data sources: physician claims, ACCS, DAD, Alberta Blue Cross
2.
Identify First Incidence
– Identify all OA - several scenarios tested different criteria
– Cross referenced those with OA from the 4 data sources
– First date of incidence identified
3.
Determine Prevalence
– Linked to AHW registry data
• Using the incident cases from step 2, can determine prevalence
• Prevalence rate = prevalent cases / total registry population
4.
Determine incidence rates
– Linked to AHW registry data
• Incident cases / (total mid-year registry population– previous year’s prevalence )
• Incident rates calculated by age and gender and year
Range of OA Incidence Estimates
Age Standardized Incidence rates (for 2008)
20
per 1000 population
16.9
15
10
8.6
8.2
7.6
8.3
8.2
7.7
5
0
Extreme
Basic
Basic + ExclBasic + Excl Basic + all Basic + Claims only
& 5 yrs exclusions ABC & all
excl
Range of OA Prevalence Estimates
160
140
120
100
80
60
Prevalence nearly
plateaus at about 8%
after 15 years of data
40
20
0
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Extreme
Basic + Excl & 5 yrs
Basic
Basic + all exclusions
Basic + Excl
Basic + Blue Cross & all excl
Prevalence per 1000 population
Example of Questions that Health Policy
Makers Would like to Address
• What resources are needed to:
• Meet the target wait times for hip and knee surgery?
• Meet current and future demand for OA services?
• What effect do acute and sub-acute utilization
policies have on system performance?
• What are the resource requirements (human,
financial, infrastructure) and how are they best
deployed to achieve expected outcomes?
• What are the wait time and cost implications of
changes in service demand and supply for OA?
Scenario 1 - Access
What if a prevention program were initiated that
reduced OA incidence by 10%?
What would the effect be on…
The number of OA patients in primary care?
 The annual number of hip and knee joint
replacement surgeries?

17
Scenario 1 - Access
32,500
fewer
Patients in Primary Care
350000
12000
300000
10000
250000
Number of Procedures
Number of Patients
Annual OA Hip & Knee Replacement
Surgeries
200000
150000
100000
8000
1,000
fewer
6000
4000
2000
50000
0
2011
2016
base case
2021
2026
incidence -10%
2031
0
2011
2016
base case
2021
2026
2031
incidence -10%
18
Scenario 2 - Effectiveness
How does resource use differ among
discharge destinations?
What is the effect on….
– Physician visit costs?
19
Scenario 2 - Effectiveness
Costs of physician resource use by facility type, 2011
$700,000
Rheumatologist
Radiologist
Physiatrist
Orthopaedic Surgeon
Internal Medicine
GP
Emergency physician
$600,000
$500,000
$400,000
$300,000
$200,000
$100,000
Calgary
Central
Edmonton
North
Sub-acute care
Other Facility
Homecare
Home
Sub-acute care
Other Facility
Homecare
Home
Sub-acute care
Other Facility
Homecare
Home
Sub-acute care
Other Facility
Homecare
Home
Sub-acute care
Other Facility
Homecare
Home
$0
South
20
Scenario 3 - Efficiency
What if all new OA patients had to visit a
screener before being eligible for a consult
with an orthopaedic surgeon?
What would the effect be on…
– Patient flows leading to surgery?
21
Scenario 3 - Efficiency
Number of Patients
2011
2035
45000
45000
40000
40000
35000
35000
30000
30000
25000
25000
20000
20000
15000
15000
10000
10000
5000
5000
0
0
Awaiting In screening Awaiting
screening
surgeon
consult
base case
In surgeon In extended Awaiting
care
pre-surgical Surgery
care
all screened
Awaiting In screening Awaiting
screening
surgeon
consult
base case
In surgeon In extended Awaiting
care
pre-surgical Surgery
care
all screened
22
Implementation as the Meeting
Ground of Decision Makers
• Clinical Care Delivery - AHS Strategic Clinical Networks:
– Establish quality standards, pathways and evidence-informed care
models and interventions to create high performing health system
• System tools will be needed to meet AHS SCN
mandate to transform OA health service delivery
• Health Policy - AHW Decision Makers:
– Informing long term health care strategies about OA prevention,
management and treatment
– Inform human resources and reimbursement policies
• System tools to explore alternative scenarios and
help inform long term strategies
Questions for Discussion (1)
• Would a system such as this be useful in guiding surgical
scheduling practices in the clinic?
• Would practitioners be willing use this tool, and if not, why?
• What additional features would make it more useful?
• What assumptions have we made that may not work in
current clinical practice?
• Is there a strong desire to provide more scheduling certainty
to elective patients?
• What are the biggest barriers to providing a surgical ‘window
when a decision to treat is made?
• What other specialties might benefit from a similar approach
or tool?
Questions for Discussion (2)
• What policy questions about osteoarthritis
management are of interest to you in your setting?
• How can the system dynamics model help inform
decisions in your setting? What scenarios are
relevant?
• What are the expected ‘bottlenecks’ in your system
of care delivery for osteoarthritis?
• Are there important aspects of osteoarthritis care
delivery missing in the osteoarthritis simulation?
If so, what are they?
Thank-you!