System Dynamics and the Physics of Possibility in Health Policy

1
System Dynamics and the Physics of Possibility
in Health Policy

• Tools for Developing a Dynamic Understanding of
  Access-to-Care Options During HIV Prevention
  Trials

Jack Homer Homer Consulting Voorhees, New Jersey
Bobby Milstein Centers for Disease Control and
PreventionAtlanta, Georgia
Family Health International October 25,
2004 Chapel Hill, NC
2
The Dynamic Dilemma of HIV Prevention Trials
3
(No Transcript)
4
When we attribute behavior to people rather than
system structure the focus of management becomes
scapegoating and blame rather than the design of
organizations in which ordinary people can
achieve extraordinary results.
Beyond Scapegoating
-- John Sterman
The tendency to blame other people instead of
the system is so strong that psychologists call
it the fundamental attribution error.
Sterman J. System dynamics modeling tools for
learning in a complex world. California
Management Review 200143(4)8-25.
5
Tools for Policy Analysis
Events
Time Series Models Describe trends

• Increasing
• Depth of causal theory
• Degrees of uncertainty
• Robustness for longer-term projection
• Value for developing policy insights

Multivariate Stat Models Identify historical
trend drivers and correlates
Patterns
Dynamic Models Anticipate future trends, and
find policies that maximize chances of a
desirable path
Structure
Developed by Jack Homer, Homer Consulting
6
Different Modeling Approaches For Different
Purposes
Logic Models (flowcharts, maps or diagrams) System Dynamics (causal loop diagrams and simulation models) Forecasting Models
Articulate steps between actions and anticipated effects Improve understanding about the plausible effects of a policy over time Focus on patterns of change over time (e.g., long delays, worse before better) Make accurate forecasts of key variables Focus on precision of point predictions and confidence intervals
7
Questions for Today

• What are system dynamics models?
• What questions guide their development?
• What the analytic steps involved?
• How can they be used to support learning and
  effective, transformative action?
• How can we begin thinking about the dynamic
  forces that affect HIV prevention trials?

8
System Dynamics Focuses on the Connection
Between Behavior and Structure
System behavior is determined by feedback
structure -- including accumulation, delay, and
nonlinear response
9
Water Glass Model Diagram (Vensim software)
10
Re-Directing the Course of ChangeQuestions from
System Modeling and Social Navigation
Where?
How?
Why?
Who?
2020
2010
11
Steps for Developing Dynamic Policy Models
Identify a Persistent Problem One that exists
due to dynamic complexity
12
Diabetes System Modeling ProjectWhere is the
Leverage for Health Protection?
People with
People with
People with
Undiagnosed,
Undiagnosed,
Undiagnosed
Uncomplicated
Complicated
PreDiabetes
Diabetes
Diabetes
People with
Normal
Glycemic
Levels
People with
People with
People with
Diagnosed,
Diagnosed,
Diagnosed
Uncomplicated
Complicated
PreDiabetes
Diabetes
Diabetes
Homer J, Jones A, Seville D, Essien J, Milstein
B, Murphy D. The CDC diabetes system modeling
project developing a new tool for chronic
disease prevention and control. 22nd
International Conference of the System Dynamics
Society Oxford, England 2004.
13
Diabetes System Modeling ProjectWhere is the
Leverage for Health Protection?
PreDiabetes
Diabetes
Detection
Detection
Developing
Diabetes from Undx
Developing
PreDiabetes
Complications from
PreD,
People with
People with
Dying from Undx
Onset
Undx diab
People with
Undiagnosed,
Undiagnosed,
Complications
Undiagnosed
Uncomplicated
Complicated
PreDiabetes
Diabetes
Diabetes
Recovering from
People with
PreDiabetes
Normal
Diagnosing
Diagnosing
Diagnosing
Uncomplicated
Glycemic
Complicated
PreDiabetes
Diabetes
Diabetes
Levels
Developing
People with
People with
Dying from
Complications
People with
Diagnosed,
Diagnosed,
Complications
Recovering from
Diagnosed
Uncomplicated
Complicated
PreDiabetes
Diabetes
PreDiabetes
Diabetes
Diabetes
Onset
Risk for
PreDiabetes
Diabetes
PreDiabetes
Control
Control
Obese Fraction of
the Population
14
Selected Data Sources for Model Calibration

• High Risk Population, Incidence, Prevalence,
  Deaths
• National Diabetes Statistics http//diabetes.nidd
  k.nih.gov/dm/pubs/statistics/index.htm
• Prevalence of Selected Chronic Conditions United
  States, 1990-1992 http//www.cdc.gov/nchs/data/s
  eries/sr_10/sr10_194.pdf
• Healthy People 2000 Review, 1997
  http//www.cdc.gov/nchs/data/hp2000/hp2k97.pdf
• Deaths Preliminary Data for 2000
  http//www.cdc.gov/nchs/data/nvsr/nvsr49/nvsr49_12
  .pdf
• Estimated number of adults with prediabetes in
  the U.S. in 2000 opportunities for prevention,
  Benjamin SM et al (DDT/CDC), Diabetes Care 26
  645-9, 2003.
• A Dynamic Markov Model for Forecasting Diabetes
  Prevalence in the United States through 2050,
  Honeycut AA et al. (DDT/CDC), Health Care Mgmt
  Sci 6 155-164, 2003.
• Complications and Benefits of Control
• Model of Complications of NIDDM--1. Model
  Construction and Assumptions, Eastman RC et al,
  Diabetes Care 20 725-734, 1997.
• Model of Complications of NIDDM--2. Analysis of
  the Health Benefits and Cost-Effectiveness of
  Treating NIDDM with the Goal of Normoglycemia,
  Eastman RC et al., Diabetes Care 20 735-744,
  1997.
• The Prevention or Delay of Type 2 Diabetes,
  position statement from ADA and NIDDK, Diabetes
  Care 25 742-749, 2002
• Effect of Improved Glycemic Control on Health
  Care Costs and Utilization, EH Wagner et al.,
  JAMA 285 182-189, 2001
• Health Economic Benefits and Quality of Life
  During Improved Glycemic Control in Patients with
  Type 2 Diabetes Mellitus A Randomized,
  Controlled Double-Blind Trial, Testa MA and
  Simonson DC, JAMA, 280 1490-6, 1998

One immediate benefit of the modeling process is
often knowledge integration
15
Diabetes System Modeling ProjectSimulating
Policy Scenarios
Historical Calibration
Exploring Plausible Futures
Diagnosed Diabetes of Adults
Diabetes-related death rate per year for adult
population
Status Quo
Disease Mgmt
Reduced Obesity
Partial Disease Mgmt Obesity Reduction
Obese of Adults
Homer J, Jones A, Seville D, Essien J, Milstein
B, Murphy D. The CDC diabetes system modeling
project developing a new tool for chronic
disease prevention and control. 22nd
International Conference of the System Dynamics
Society Oxford, England 2004.
16
Setting Realistic ExpectationsHP 2010 Diabetes
Objectives
Baseline HP 2010 Target Percent Change
Reduce Diabetesrelated Deaths Among Diagnosed (5-6) 8.8 per 1,000 7.8 -11
Increase Diabetes Diagnosis (5-4) 68 80 18
Reduce New Cases of Diabetes (5-2) 3.5per 1,000 2.5 -29
Reduce Prevalence of Diagnosed Diabetes (5-3) 40 per 1,000 25 -38
U.S. Department of Health and Human Services.
Healthy People 2010. Washington DC Office of
Disease Prevention and Health Promotion, U.S.
Department of Health and Human Services 2000.
http//www.healthypeople.gov/Document/HTML/Volume1
/05Diabetes.htm
17
The Simple Physics of Diabetes
People with
Undiagnosed
Initial
Diabetes
Onset
With a diagnosed onset flow of 1.1 mill/yr
Diagnosed
Onset
People with
Diagnosed
Dying from Diabetes
Diabetes
Complications
It is impossible for any policy to reduce
prevalence38 by 2010!
And a death flow of 0.5 mill/yr (4/yr rate)
18
History and Futures for Diabetes
PrevalenceReported Trends, HP Objectives, and
Simulation Results
Reported
Meet Detection Objective (5-4)
I
Status Quo
G
Meet Onset Objective (5-2)
H
F
D
C
HP 2000 Objective
HP 2010 Objective (5-3)
E
19
Projecting the Community-Wide Costs and Benefits
of Pursuing Perfection in Whatcom County, WA
Jack Homer Gary Hirsch
20
Whatcom County Pursuing Perfection (P2) Program

• A patient-centered team approach supported by
• Electronically shared clinical information
  medical record, care plan, medication list
• Idealized design of clinical office practice
  (IDCOP) for greater access and efficiency
• Evidence-based guidelines
• A clinical care specialist (RN) when needed
• Cost-effective screening preventive measures
• Initial disease focus diabetes, heart failure
• Initial community participants family practice
  group, cardiology group, geriatric practice, a
  community health center, the hospital, and three
  insurers
• Two years of funding by Robert Wood Johnson
  Foundation

21
P2 Modeling Framework
Program Adoption
Program Personnel Info System Costs
Healthcare Utilization
Total Costs to Program, Payors, Providers,
Patients, Employers
Personnel include administrators and staff,
process/OD consultants, and clinical care
specialists.
22
Deaths from Diabetes 2001-21 Four Scenarios
Disease-related deaths per year
Status Quo
Disease management only
Full program
Full program Medicare drug coverage
Full program includes community-based
screening positives are referred to physician
for follow-up testing and counseling.
A similar pattern of results is seen for
diabetes-related disability losses.
23
Impact of Full Program on Spending for Diabetes
and Heart Failure 2001-21
Constant (2001) dollars per year
Even during 2003-08 period of increased spending,
cost per dollar of disability loss avoided
averages only 0.37, and cost per life-year saved
only 22,000.
Including Infrastructure Costs
Excluding Infrastructure Costs (Health care
spending only)
Health care spending by insurers and patients
pays for physicians, hospital, ancillary
services, hospice, home care, skilled nursing
facility, exercise rehab, drugs, and implanted
devices.
24
Financial Impacts 2003-08 Winners and
Losers (in Year 2001 dollars)
25
Commitment to a Comprehensive Strategy (based on
model sensitivity testing)

• Disease management quickly starts improving
  health outcomes, but does not by itself reduce
  total spending
• Preventive measures produce increasing savings
  over time
• Solid program execution that delivers expected
  health benefits is necessary to achieve savings
• Clinical care specialists must be sufficient to
  meet referral demand
• Full drug coverage for the elderly would further
  improve health outcomes and program
  cost-effectiveness

26
Toward a Dynamic View of HIV Prevention Trials
27
Toward a Dynamic View of HIV Prevention Trials
28
Toward a Dynamic View of HIV Prevention Trials
29
Toward a Dynamic View of HIV Prevention Trials
30
Toward a Dynamic View of HIV Prevention Trials
31
Toward a Dynamic View of HIV Prevention Trials
32
Toward a Dynamic View of HIV Prevention Trials
33
Toward a Dynamic View of HIV Prevention Trials
34
Widespread Interest in the Promise of a Systems
Orientation
See http//www.cdc.gov/syndemics/ajph-systems.htm
Submission Deadline February 1, 2005