System Dynamics: Projecting the Community-Wide Costs and Benefits of

1
System Dynamics Projecting the Community-Wide
Costs and Benefits of Pursuing Perfection in
Whatcom County, WA
Jack Homer Gary Hirsch Independent Consultants
to PeaceHealth American Hospital
Association Washington, DC May 9, 2003
2
P2 in Whatcom County

• Individual patient goals, values and needs are
  the central focus.
• Across the entire community build safety,
  timeliness, effectiveness, efficiency, and
  equality into the health care system.
• Support each patient and their care team with
• A shared care plan
• A shared medication list
• Access to clinical information at all times
• Idealized design of clinical office practice
  (IDCOP), including group visits and telephone and
  e-mail consults
• Evidence-based guidelines
• A clinical care specialist when needed
• Promote cost-effective screening, preventive
  education, and risk management

3
(No Transcript)
4
Achieving the Vision

• Initial disease focus Diabetes and Heart Failure
  
• Initial community participants A few medical
  groups, the hospital, and one insurer
• Two years of funding by RWJF
• Networking with other like minded communities in
  the US and Europe Pursuing Perfection partners

5
Reasons for Projecting Program Impacts

• Resource Planning
• Fixed costs to program and providers

• Other Financial Needs and Concerns
• Winners and losers redistribution schemes

• Critical Success Factors
• What is critical and what is not?

• Set Expectations
• Anticipate pain that precedes gain prevent
  frustrations
• Recognize when adjustments are needed and when
  not

6
System Dynamics Modeling Indirect Impacts, Too

7
Framework for Projecting Program Impacts for a
Single Chronic Illness
8
Type 2 Diabetes Progression Care
6-7 of the adult population is diabetic,
including 17 of the elderly. 35-40 of
diabetics are undiagnosed Stage 1. Of the
diagnosed, about 50 are Stage 1, 30 Stage 2,
20 Stage 3, and about 40 have their blood sugar
under control.
9
Heart Failure (HF) Progression Care
3 of the adult population has HF, including 16
of the elderly. 50-55 of HF is asymptomatic
(Stage B). Only 20 of asymptomatic HF is
diagnosed. Of those with symptomatic HF (Stages C
and D), 15-20 have been hospitalized within the
past year, with an annualized death rate of
30-35.
10
Data Sources

• State Office of Fiscal Management
• Population projections
• National Center for Health Statistics (CDC)
• Disease prevalence
• Group Health Cooperative
• Disease prevalence and stage distribution
• Healthcare utilization and costs
• St. Joseph Hospital and Local Physician Practices
• Healthcare utilization and costs
• Medical Literature and Clinician Estimates
• Disease prevalence and stage distribution
• Healthcare utilization and costs
• Disability days and costs
• Benefits and costs of ideal versus usual care

11
Growth of Diabetes in Whatcom County 2001-21
(Status Quo Projection)
Average growth 2.2 per year over 20
years. Prevalence rises from 6.5 to 7.5 of
adult population.
Undiagnosed Stage 1
Stage 3 (Post Organ Failure)
Stage 2 (Organ Disease)
Diagnosed Stage 1
12
Diabetes-Related Costs in Whatcom County 2001-21
(Status Quo Projection)
In constant (2001) healthcare dollars per year
Social loss
Employer loss
Pharmacy
Provider Revenue Ancillary
Provider revenue here includes all visits,
whether clearly diabetes-related or not.
13
Growth of Heart Failure in Whatcom County 2001-21
(Status Quo Projection)
Average growth 3.6 per year over 20
years. Prevalence rises from 3.0 to 4.5 of
adult population.
Undiagnosed Stage B
Stage D (Recently Hospitalized)
Stage C (Ever Symptomatic)
Diagnosed Stage B (Never symptomatic)
14
HF-Related Costs in Whatcom County 2001-21
(Status Quo Projection)
In constant (2001) healthcare dollars per year
Social loss
Employer loss
Implanted devices
Pharmacy
SNF, Hospice, Home care, Rehab
Provider Revenue Ancillary
Provider revenue here includes only those
admissions and visits (about 39 of the total
for HF patients) clearly related to heart failure.
15
Program Elements as Modeled

• Program adoption by MDs starts 2002, reaches
  100 by 2005
• Participating MDs pay for clinical information
  systems
• Greater intensity of planned primary care,
  including group visits and telephone and e-mail
  consults
• Greater use of exercise rehab for symptomatic HF
  patients
• Greater use of drugs inadequate drug coverage a
  possible hindrance
• Program personnel
• Fixed administrative costs incurred starting 2001
• Process/OD consultants assist MD office redesign
  2002-2007
• Clinical care specialists hired to meet
  model-projected demand
• Year 01-02 03 04 05-10
  11-13 14-18 19-21
• CCS FTEs 2 3 5
  7 8 9 10
• Community-based diabetes screening and preventive
  ed starts 2004
• 8/subject plus positive reading generates
  additional visit to PCP

16
Program Infrastructure Costs 2001-21 (in Year
2001 dollars per year)
Displayed values are for 2008.
42K
Diabetes Screening
Clinical Info Systems
438K
793K
Personnel (Admin, Process/OD, CCSs)
17
Program Benefit Assumptions
18
Deaths from Diabetes HF 2001-21
Disease-related deaths per year
Status Quo
361
With Program
247
Displayed values are for 2008.
19
Disability Days from Diabetes HF 2001-21
Disease-related disability weekdays per year
Status Quo
219 thou
With Program
166 thou
Displayed values are for 2008.
20
Healthcare Spending for Diabetes HF 2001-21
In constant Year 2001 dollars per year
Status Quo
86.3M
With Program
86.0M
Displayed values are for 2008.
Includes payments by insurers and patients for
Physicians, Hospital, Ancillary services,
Hospice, Home care, Skilled nursing facility,
Exercise rehab, Drugs, and Implanted devices.
21
Overall Program Costs Benefits 2003-08
Cumulative (in Year 2001 dollars)





Disability losses Based on literature, assume
employed person (typically under 65) costs
employers an average of 120 per disability day
and costs society another 116 while an
unemployed person costs society an average of 74
per disability day.
22
Critical Success Factors (based on model
sensitivity testing)

• Disease management quickly starts improving
  health outcomes, but does not by itself reduce
  total spending (healthcare plus program
  infrastructure) below status quo

• Preventive measures (screening, preventive ed,
  risk management) generate increasing savings over
  time, and make it possible to reduce total
  spending below status quo

• Total spending reduction will not be achieved if
  program benefits are at low end of spectrum
  e.g., for reason of lax program execution

• Clinical care specialists must be sufficient to
  meet referral demand, else program
  cost-effectiveness is compromised

• Comprehensive drug coverage for the elderly would
  improve health outcomes further, and further
  boost program cost-effectiveness

• Savings can still be achieved, though more
  modestly, even if less than 100 of communitys
  MDs participate in program

23
Program Financial Impacts 2003-08 Winners and
Losers (in Year 2001 dollars)
24
Financial Needs and Concerns

• Program personnel Ongoing requirement of up to
  1 million/year

• Diabetes screening and prevention education
  relatively inexpensive

• Clinical information systems about 1,500 per
  physician per year

• Practice redesign (IDCOP) some MD costs during
  phase-in period

• Hospital loss of income relative to status quo
  exceeds 2 million per year by 2007 concern
  about internally-subsidized programs (e.g.,
  mental and behavioral health services)

• Increase in patient self-payments (primarily
  drugs) of about 150 per diabetic and 600 per HF
  patient, assuming no improvement in drug coverage
  for the elderly

25
Payment Mechanism Experiment Using Capitation to
Smooth Impact on Hospital

• Problem Success at controlling diabetes reduces
  hospital income, possibly jeopardizing subsidized
  programs (mental health, substance abuse)
• Examine alternative Medicare payment scheme for
  hospital
• No fee for service
• Risk-adjusted capitation payments for patients at
  different stages covering all inpatient and
  outpatient care
• Stage 1 diabetic 2,500 per pt per yr
• Stage 2 diabetic 5,000 per pt per yr
• Stage 3 diabetic 12,500 per pt per yr
• Can hospital be made whole without requiring
  increased Medicare payments?

26
Risk-Adjusted Capitation Maintains Hospital
Income at Status Quo Level
In constant Year 2001 dollars per year
Status Quo
Program with Capitation
Program with FFS
27
Medicare Pays Only What It Would Have Paid Under
Status Quo
In constant Year 2001 dollars per year
Status Quo
Program with Capitation
Program with FFS
28
Rx Drug Coverage Experiment Full Coverage for
Medicare Patients

• Problem Many seniors, esp. those lacking Rx
  drug coverage, may be unable or unwilling to pay
  additional for drugs to achieve control
• Test Increase fraction of Medicare patients
  with drug coverage from 41 to 100, and assume
  minimal co-pay affordable by all
• Will increased pharmacy costs pay off
  sufficiently in terms of improved health and
  reduced urgent care?

29
Full Coverage Increases Drug Costs for Medicare
Patients by 1 Million per Year
In constant Year 2001 dollars per year
Program plus Full Medicare Coverage
Program with Current Coverage
Status Quo
30
But By Helping More Medicare Patients Get Their
Diabetes Under Control...
Program plus Full Medicare Drug Coverage
Program with Current Coverage
Status Quo
31
Improved Drug Coverage Helps to Further Reduce
Deaths...
Status Quo
Program with Current Coverage
Program plus Full Medicare Drug Coverage
32
And Further Reduces Disability Costs...
In constant Year 2001 dollars per year
Status Quo
Program with Current Coverage
Program plus Full Medicare Drug Coverage
33
And Actually Reduces Medicare Payments by 1
Million per Year on Net, Because of Fewer Disease
Complications
In constant Year 2001 dollars per year
Status Quo
Program with Current Coverage
Program plus Full Medicare Drug Coverage
34
Addressing Interactions Among Chronic Illnesses
(Cardiovascular Disease Example)
One type of illness may be a risk factor for
other types. Although we may ultimately want to
build multi-illness models to explore such
linkages fully, it is possible to recognize
linkages in a single-illness model. For example,
in our model of heart failure, risk management
implies better care of hypertension and high
cholesterol. But because such improved care
reduces the risk of not only heart failure but
other illnesses as well, we assign only some
fraction of the cost of such care to heart
failure per se.
35
A Feedback Diagram Depicting Quality Chasm
Issues and Their Interactions
36
Other Scenarios We Have Tested

• SET 1
• Partial program adoption by MDs
• FCN, SeaMar, and CSH adopt, but other PCPs do not
• Partial adoption affects disease and risk
  management, but diabetes screening and preventive
  ed are unaffected
• Full adoption plus comprehensive Medicare drug
  coverage
• SET 2
• Full adoption but disease management only
• Diabetes No community-based screening or
  preventive ed
• Heart failure No risk management beyond status
  quo 25
• Full adoption but with benefits undercut by poor
  execution
• Less effective preventive ed (Diabetes) or risk
  management (HF) leading to less reduction of
  disease incidence rates
• Less effective disease management leading to less
  reduction of disease progression and complication
  rates

For heart failure, assumed impacts of poor
execution based on least beneficial outcomes
reported in the literature on ideal care. For
diabetes, assumed impacts of poor execution based
on expert judgment.
37
Controlled Fraction of Known Diabetics
Full Adoption Drug Coverage
Full Adoption
Partial Adoption
Status Quo
Starting from a status quo of 40, it is assumed
that 80 of known diabetics could be brought
under control if they could afford the meds and
there were sufficient CCS support. With Medicare
drug coverage, a value of 77 is achieved, short
of the 80 only because of the steady influx of
newly diagnosed diabetics not yet under control.
38
Deaths from Diabetes (scenario set 1)
Status Quo
Partial adoption
Full adoption
Full adoptionDrug coverage
Greater control of diabetes translates into fewer
deaths.
39
Deaths from Diabetes (scenario set 2)
Status Quo
Full adoption but disease mgmt only
Full adoption but poor execution
Full adoption
Disease management starts reducing
diabetes-related deaths immediately, but over the
longer term screening and preventive ed are
responsible for at least one-third of the
reduction in deaths seen under Full adoption.
Poor program execution would, from the outset,
reduce the number of life-years saved by perhaps
one-third.
40
Possible Funding Sources

• A variety of possibilities for direct or indirect
  funding for the immediate future and beyond
• Foundations
• Congressional appropriation
• Pharmaceutical companies
• Community tax levy for healthcare
• Disability insurance premium reductions

• Leveraging Medicare payments
• Near term demonstration grant (3 to 5 years)
• Longer term changes in policy (e.g.,
  severity-adjusted capitation, expanded drug
  coverage)
• Even today, the community can benefit if private
  insurers offer Medicare plans and use some of the
  anticipated net profits to help fund Pursuing
  Perfection