Modeling Population Dynamics - PowerPoint PPT Presentation

Loading...

PPT – Modeling Population Dynamics PowerPoint presentation | free to download - id: 66b3f4-OThmY



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Modeling Population Dynamics

Description:

Obesity Bobby Milstein Syndemics Prevention Network ... then extend to the future Explore effects of new interventions affecting caloric balance Focusing by age, ... – PowerPoint PPT presentation

Number of Views:31
Avg rating:3.0/5.0
Slides: 54
Provided by: cdc61
Learn more at: http://www2.cdc.gov
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Modeling Population Dynamics


1
Modeling Population Dynamics
A Work in Progress Dialogue
  • Obesity

Bobby Milstein Syndemics Prevention
Network Centers for Disease Control and
Prevention Atlanta, Georgia bmilstein_at_cdc.gov
Jack Homer Homer Consulting Voorhees,
NJ jhomer_at_comcast.net
CDC Diabetes and Obesity Conference Denver,
CO May 17, 2006
2
Topics for Today
  • Dynamic modeling for learning and action
  • Structure of the current model
  • Dynamic population weight framework
  • Calibrating the model
  • Behavior of the current model
  • A status quo future
  • Alternative futures
  • Conclusions, questions, and next steps

3
Contributors
  • System Dynamics Consultants
  • Jack Homer
  • Gary Hirsch
  • Project Coordinator
  • Bobby Milstein
  • Core Design Team
  • Dave Buchner
  • Andy Dannenberg
  • Bill Dietz
  • Deb Galuska
  • Larry Grummer-Strawn
  • Anne Hadidx
  • Robin Hamre
  • Laura Kettel-Khan
  • Elizabeth Majestic
  • Jude McDivitt
  • Cynthia Ogden
  • Michael Schooley
  • Time Series Analysts
  • Danika Parchment
  • Cynthia Ogden
  • Margaret Carroll
  • Hatice Zahran
  • Workshop Participants
  • Atlanta, GA May 17-18 (N47)
  • Lansing, MI July 26-27 (N55)

4
Purposes for Modeling Obesity Dynamics Primary
Uses and Users
  • Chart Progress Toward Goals (Planners/Evaluators/M
    edia)
  • Set justifiable goals
  • Define a status quo future, as well as
    plausible alternatives based on policy scenarios
  • Estimate how strong interventions must be to make
    a difference, and how long it will take for those
    effects to become visible
  • Develop Better Measures and New Knowledge
    (Researchers)
  • Integrate diverse data sources into a single
    analytic environment
  • Infer properties of unmeasured or poorly measured
    parameters
  • Convene Multi-stakeholder Action Labs
    (Policymakers)
  • Understand how a dynamically complex obesity
    system functions
  • Discover short- and long-term consequences of
    alternative policies

5
Modeling Obesity Dynamics Opportunities to
Integrate Diverse Policy Perspectives
  • Lifecourse Perspective
  • Consider life-long impacts and intergenerational
    effects
  • Ecological Perspective
  • Consider (a) weight-related behaviors, (b)
    behavioral settings, (c) social-cultural-economic-
    political forces, and (d) other health
    conditions, all by social position
  • Action Perspective
  • Clarify how obesity can be reduced (i.e., what
    kinds of actions are needed)
  • Clarify who is in a position to take those
    actions (i.e., roles for different types of
    organizations)
  • Estimate how strong new programs/policies must be
    to make a difference, as well as when those
    effects will become visible
  • Navigational Perspective
  • Set justifiable goals for the future, given
    existing momentum
  • Chart progress (annually?) by surveying actions
    and anticipating trajectories of change

Others.
6
Re-Directing the Course of Change Questions
Addressed by System Dynamics Modeling
What?
Where?
Prevalence of Obese Adults, United States
Why?
How?
Who?
Data Source NHANES
7
Modeling for Learning and Action
Multi-stakeholder Dialogue
Plausible Futures (Policy Experiments)
Dynamic Hypothesis (Causal Structure)
  • Model Structure
  • Trace changes in caloric balance through to
    overweight and obesity prevalence1
  • Trace intervention effects over the lifecourse
    by age and sex
  • Intervention Scenarios
  • Efforts to alter caloric balance via intensive
    weight loss/maintenance services and/or via
    broad changes in peoples food and activity
    environment
  • Focusing by age range and sex
  • Focusing by BMI category

1 Because health burden is associated with the
obese tail of the BMI distribution, and cannot be
accurately estimated from mean BMI alone
8
Major Project Phases
  • Conceptualization and Data Gathering (May 2005
    July 2005)
  • Convene stakeholder workshops
  • Collect time series data
  • Develop multiple iterations of a dynamic
    hypothesis
  • Formulation, Calibration, and Testing (August
    2005 November 2005)
  • Assure appropriate fit to history
  • Examine future behavior under status quo as well
    as policy scenarios
  • Policy Scenarios and Goal-setting (December 2005
    April 2006)
  • Study major classes of interventions, alone and
    in combination
  • Learn how strong new interventions must be to
    make a lasting difference, as well as how long it
    will take for those effects to become visible
  • Further Testing (May 2006 July 2006)
  • Conduct sensitivity tests to see if data
    uncertainties affect policy conclusions
  • Elicit feedback from SD experts

9
System Dynamics Was Developed to Address
Problems Marked By Dynamic Complexity
  • Origins
  • Jay Forrester, MIT (from late 1950s)
  • Public policy applications starting late 1960s
  • Good at Capturing
  • Differences between short- and long-term
    consequences of an action
  • Time delays (e.g., transitions, detection,
    response)
  • Accumulations (e.g., prevalence, capacity)
  • Behavioral feedback (e.g., actions trigger
    reactions)
  • Nonlinear causal relationships (e.g., effect of
    X on Y is not constant-sloped)
  • Differences or inconsistencies in goals/values
    among stakeholders

Sterman JD. Business dynamics systems thinking
and modeling for a complex world. Boston, MA
Irwin McGraw-Hill, 2000. Homer JB, Hirsch GB.
System dynamics modeling for public health
background and opportunities. American Journal of
Public Health 200696(3)452-458.
10
Understanding Dynamic Complexity Longand often
surprisingchains of cause and effect
Forrester JW. Counterintuitive behavior of social
systems. Technology Review 197173(3)53-68. Meado
ws DH. Leverage points places to intervene in a
system. Sustainability Institute, 1999.
Available at lthttp//www.sustainabilityinstitute.
org/pubs/Leverage_Points.pdfgt. Richardson GP.
Feedback thought in social science and systems
theory. Philadelphia, PA University of
Pennsylvania Press, 1991. Sterman JD. Business
dynamics systems thinking and modeling for a
complex world. Boston, MA Irwin McGraw-Hill,
2000.
11
Tools for Policy Analysis
Events
Time Series Models Describe trends
  • Increasing
  • Depth of causal theory
  • Data and sensitivity testing requirements
  • Robustness for longer-term projection
  • Value for developing policy insights

Multivariate Stat Models Identify historical
trend drivers and correlates
Patterns
Dynamic Simulation Models Anticipate new
trends, learn about policy consequences, and set
justifiable goals
Structure
12
An Ecological Framework for Organizing Influences
on Overweight and Obesity
Adapted from Koplan JP, Liverman CT, Kraak VI,
editors. Preventing childhood obesity health in
the balance. Washington, DC Institute of
Medicine, National Academies Press 2005.
13
A Conventional View of Causal Forces
Wider Environment (Economy,
Health Conditions
Technology, Laws) Influence
Detracting from
Genetic Metabolic
on Healthy Diet Activity
Healthy Diet Activity
Rate Disorders
Options Available at
Prevalence of
Home, School, Work,
Healthiness of Diet
Overweight
Activity Habits
Community Influencing
Related Diseases
Healthy Diet Activity
Media Messages
Healthcare Services
Promoting Healthy
to Promote Healthy
Diet Activity
Diet Activity
14
A Conventional View of Causal Forces
  • This sort of open-loop (non-feedback) approach
  • Ignores intervention spill-over effects and often
    suggests the best strategy is a multi-pronged
    fill all needs one (even if not practical or
    affordable)
  • Ignores unintended side effects and delays that
    produce short-term vs. long-term differences in
    outcomes
  • Cannot fairly evaluate a phased approach e.g.
    bootstrapping which starts more narrowly
    targeted but then broadens and builds upon
    successes over time

15
The Rise and Future Fall of Obesity The Why and
the How in Broad Strokes
Drivers of Unhealthy Habits
Time
16
A Closed-Loop View of Causal Forces
DRAFT 5/8/05
17
A Closed-Loop View of Causal Forces
DRAFT 5/8/05
18
A Closed-Loop View of Causal Forces
DRAFT 5/8/05
19
The Closed-Loop View Leads Us To Question
  • How can the engines of growth loops (i.e. social
    and economic reinforcements) be weakened?
  • What incentives can reward parents, school
    administrators, employers, and other
    decision-makers for expanding healthy diet and
    activity options ?
  • Are there resources for health protection that do
    not compete with disease care?
  • How can industries be motivated to change the
    status quo rather than defend it?
  • How can benefits beyond weight reduction be used
    to stimulate investments in expanding healthier
    options?

20
Building a Foundation for Analysis Structure of
the Current Model
21
Obesity Prevalence Over the Decades Two Broad
Phases
Phase 1 Calculating Obesity Dynamics
Dynamic Population Weight Framework (BMI
Surveillance, Demography, and Nutritional
Science)
Policy Drivers (Trends Interventions Affecting
Caloric Balance by Age, Sex, BMI Category, etc)
Policy Drivers (Trends Interventions Affecting
Caloric Balance by Age, Sex, BMI Category, etc)
Consequences Over Time Changing Prevalence of
Four BMI Categories 1970-2050
22
Summary of Current Direction
  • Simulate overweight and obesity prevalences over
    the life-course
  • Reproduce relative stability in the 1970s and
    growth to the present, then extend to the future
  • Explore effects of new interventions affecting
    caloric balance
  • Focusing by age, sex, and/or BMI category
  • Treat intervention details (composition,
    response, coverage, efficacy, cost) as exogenous
  • Not yet addressing feedback loops of
    reinforcement and resistance
  • Not yet addressing cost-effectiveness

23
Obesity Dynamics Over the Decades Dynamic
Population Weight Framework
24
Obesity Dynamics Over the Decades Dynamic
Population Weight Framework
25
BMI Category Definitions
  • For infants (ages 0-23 months)
  • Not overweight weight-for-recumbent length
    (WRL)lt85th percentile
  • Moderately overweight WRLgt85th percentile and
    lt95th percentile
  • Moderately obese WRLgt95th percentile and lt99th
    percentile
  • Severely obese WRLgt99th percentile
  • For youth (ages 2-19)
  • Not overweight BMIlt85th percentile or 25
  • Moderately overweight BMIgt85th percentile and
    25 and lt95th percentile or 30
  • Moderately obese BMIgt95th percentile and 30
    and lt99th percentile or 35
  • Severely obese BMIgt99th percentile and 35
  • For adults (ages 20)
  • Not overweight BMIlt 25
  • Moderately overweight BMIgt25 and lt30
  • Moderately obese BMIgt30 and lt35
  • Severely obese BMIgt35

Percentiles from CDC Growth Charts based on
NHANES I and II measurements.
26
Obesity Dynamics Over the Decades Dynamic
Population Weight Framework
Indicates possible extensions to the existing
model
27
Obesity Dynamics Over the Decades Dynamic
Population Weight Framework
Indicates possible extensions to the existing
model
28
Obesity Prevalence Over the Decades
Dynamic Population Weight Framework
29
Obesity Dynamics Over the Decades Drivers of
Change
Indicates possible extensions to the existing
model
30
Obesity Dynamics Over the Decades Drivers of
Change
Indicates possible extensions to the existing
model
31
Obesity Dynamics Over the Decades Drivers of
Change
Indicates possible extensions to the existing
model
32
Obesity Dynamics Over the Decades Drivers of
Change
Indicates possible extensions to the existing
model
33
Obesity Dynamics Over the Decades Drivers of
Change
Indicates possible extensions to the existing
model
34
Calibrating the Model Estimating Flow-Rates and
Past Changes in Caloric Balance
35
Information Sources
Topic Area Data Source
Prevalence of Overweight and Obesity Prevalence of Overweight and Obesity
BMI prevalence by sex and age (10 age ranges) National Health and Nutrition Examination Survey (1971-2002)
Translating Caloric Balances into BMI Flow-Rates Translating Caloric Balances into BMI Flow-Rates
BMI category cut-points for children and adolescents CDC Growth Charts
Median BMI within each BMI category National Health and Nutrition Examination Survey (1971-2002)
Median height National Health and Nutrition Examination Survey (1971-2002)
Average kilocalories per kilogram of weight change Forbes 1986
Estimating BMI Category Down-Flow Rates Estimating BMI Category Down-Flow Rates
In adults Self-reported 1-year weight change by sex and age NHANES (2001-2002) indicates 7-12 per year
In children BMI category changes by grade and starting BMI Arkansas pre-K through 12th grade assessment (2004-2005) indicates 15-28 per year
Population Composition Population Composition
Population by sex and age U.S. Census and Vital Statistics (1970-2000 and projected)
Death rates by sex and age U.S. Census and Vital Statistics (1970-2000 and projected)
Birth and immigration rates U.S. Census and Vital Statistics (1970-2000 and projected)
Influence of BMI on Mortality Influence of BMI on Mortality
Impact of BMI category on death rates by age Flegal, Graubard, et al. 2005.
36
Data Uncertainties Limitations
  • No reliable longitudinal data on caloric intake
    and expenditure broken out by age, sex, BMI
    category
  • Reliable NHANES data on blacks and
    Mexican-Americans only since NHANES III (1988-94)
  • NHANES prevalence estimates are imprecise
  • May affect timing of inferred growth inflection
    point
  • Down-flow rate constants are imprecise
  • Dont know to what extent historical caloric
    imbalances have led to increase in up-flows as
    opposed to decrease in down-flows
  • We have assumed entirely the former

37
Growth of Obesity for Four Age Ranges 1960-2002
Definitions Ages 2-19 (NHES) Obese
BMIgt95th percentile on CDC growth chart Ages
2-19 (NHANES) Obese BMIgt30 or gt95th
percentile on CDC growth chart Ages 20-74
Obese BMIgt30
38
Growth of Obesity for Four Age Ranges 1960-2002
Definitions Ages 2-19 (NHES) Overweight
BMIgt85th percentile, Obese BMIgt95th percentile
on CDC growth chart Ages 2-19 (NHANES)
Overweight BMIgt25 or 85th percentile, Obese
BMIgt30 or 95th percentile,
Severely obese BMIgt35 or 99th percentile on CDC
growth chart Ages 20-74
Overweight BMIgt25 Obese BMIgt30 Severely obese
BMIgt35
39
Calibration of Uncertain Parameters To Reproduce
60 BMI Prevalence Time Series (10 age ranges x 2
sexes x 3 high-BMI categories)
  • Step 1 Adjust uncertain constants and initial
    values to get near steady-state BMI prevalence
    for the early 1970s
  • In this step, assume no change in caloric balance
    after 1970
  • Adjust 1970 up-rates and down-rates so that BMI
    prevalences settle-out at historical 1970s
    values
  • Set 1970 BMI prevalences (by annual age) to
    settled-out values
  • Repeat/adjust as necessary to minimize number of
    peaks and valleys (with increasing age) in
    assumed 1970 BMI prevalences
  • Step 2 Adjust uncertain time series inputs to
    reproduce BMI prevalence growth patterns for the
    1980s and 1990s
  • To explain increasing overweight in infants, must
    assume increasing overweight/obesity at birth (3
    series)
  • For non-infants, adjust caloric balances (54
    series by age, sex, and for Not Overwt, Mod
    Overwt, and Obese) to reproduce BMI growth
  • Calibrate from youngest age range to oldest
  • Within each age range calibrate first Overweight,
    then Obese, then Severely obese

40
Translating Caloric Balance Changes (?K) into
Flow Rate Changes (?F)
  • Parameters (for each age range and sex)
  • Cut-points for BMI categories (bc)
  • Median BMI within each BMI category (bm)
  • Median height (hm)
  • Assumption for the average number of kilocalories
    per kilogram of weight change (k)
  • Forbes empirical estimate of 8,050 kcal./kg
  • Implicitly takes into account the efficiency of
    weight deposition reflecting metabolic and other
    regulatory adjustments.
  • Glosses over known differences among individuals
    starting weight, composition of diet, efficiency
    of weight deposition

Forbes GB. Human body composition growth, aging,
nutrition, and activity. Springer Berlin,
Heidelberg 1987. Forbes GB. Deliberate
overfeeding in women and men Energy costs and
composition of the weight gain. British Journal
of Nutrition 561-9 1986.
41
Reproducing Historical Data One of 20 sex, age
Subgroups Females age 55-64
(a) Overweight fraction
(b) Obese fraction
80
50
40
60
30
Fraction of women age 55-64
Fraction of women age 55-64
40
20
20
10
0
0
1970
1975
1980
1985
1990
1995
2000
2005
1970
1975
1980
1985
1990
1995
2000
2005
NHANES
Simulated
NHANES
Simulated
(c) Severely obese fraction
25
20
15
Fraction of women age 55-64
10
5
0
1970
1975
1980
1985
1990
1995
2000
2005
NHANES
Simulated
Note S-shaped curves, with inflection in the
1990s
42
Explaining BMI Prevalence Growth Age-to-Age
Carryover Caloric Imbalance Example Females
Age 55-64
Severely obese fractions of middle-aged women
Overweight fractions of middle-aged women
Obese fractions of middle-aged women
25
80
50
20
40
60
15
30
Fraction of women by age group
Fraction of women by age group
Fraction of women by age group
40
10
20
20
5
10
0
0
0
1970
1975
1980
1985
1990
1995
2000
2005
1970
1975
1980
1985
1990
1995
2000
2005
1970
1975
1980
1985
1990
1995
2000
2005
Age 55-64
Age 45-54
Age 55-64
Age 45-54
Age 55-64
Age 45-54
Estimated caloric imbalances for women age 55-64
20
15
Kcal per day
10
5
0
1970
1975
1980
1985
1990
1995
2000
2005
Not overwt
Mod overwt
Obese
43
Estimated Caloric Balances in 1990 and 2000 For
Every Age Range BMI Category (vs. 1970)
44
Behavior of the Current Model
45
Assumptions for Future Scenarios
  • Base Case
  • Caloric balances stay at 2000 values through 2050
  • Altering Food and Activity Environments
  • Efforts to reduce caloric balances to their 1970
    values by 2015
  • Focused on
  • School Youth youth ages 6-19
  • All Youth all youth ages 0-19
  • SchoolParents school youth plus their parents
  • Used 2000 Census birth data by age of mother to
    estimate of each adult age range that are
    parents of 6-19 year olds
  • All Adults all adults ages 20
  • All Ages all youth and adults
  • Subsidized Weight Loss Programs for Obese
    Individuals
  • Net daily caloric reduction of program is 40
    kcal/day (i.e., 14,600 kcal/year or 1.8kg weight
    loss per year)
  • Fully effective by 2010 and terminated by 2020
  • All AgesWtLoss program applies to all obese
    youth and adults, and occurs on top of the All
    Ages environmental improvement scenario

46
Exploring Future Scenarios Through Simulation
Experiments
Intervention Scenario Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Weight Loss Programs for Obese Individuals Selected Results Selected Results Selected Results Selected Results
Intervention Scenario Pre-School School-age Youth Adult Parents of School-aged Youth All Other Adults All Ages Obese Fraction Among Teens (12-19) Obese Fraction Among Teens (12-19) Obese Fraction Among Adults (20-74) Obese Fraction Among Adults (20-74)
Intervention Scenario Pre-School School-age Youth Adult Parents of School-aged Youth All Other Adults All Ages 2020 2050 2020 2050
Base or Status Quo -- -- -- -- --
School Youth ü
All Youth ü ü
School Parents ü ü
All Adults ü ü
All Ages ü ü ü ü
All Ages Wt Loss ü ü ü ü ü
47
Alternative Futures Obesity in Teens (12-19)
Obese fraction of Teens (Ages 12-19)
50
40
30
Fraction of popn 12-19
20
10
0
1970
1980
1990
2000
2010
2020
2030
2040
2050
Base
SchoolYouth
AllYouth
AllAgesWtLoss
48
Alternative Futures Obesity in Adults (20-74)
Obese fraction of Adults (Ages 20-74)
50
40
30
Fraction of popn 20-74
20
10
0
1970
1980
1990
2000
2010
2020
2030
2040
2050
Base
SchoolYouth
AllYouth
SchoolParents
AllAdults
AllAges
AllAgesWtLoss
49
Exploring Future Scenarios Through Simulation
Experiments
Intervention Scenario Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Changing Food Activity Environments Focused On Weight Loss Programs for Obese Individuals Selected Results Selected Results Selected Results Selected Results
Intervention Scenario Pre-School School-age Youth Adult Parents of School-aged Youth All Other Adults All Ages Obese Fraction Among Teens (12-19) Obese Fraction Among Teens (12-19) Obese Fraction Among Adults (20-74) Obese Fraction Among Adults (20-74)
Intervention Scenario Pre-School School-age Youth Adult Parents of School-aged Youth All Other Adults All Ages 2020 2050 2020 2050
Base or Status Quo -- -- -- -- -- 20.1 20.0 37.9 39.1
School Youth ü 11.5 10.1 37.3 36.6
All Youth ü ü 9.7 6.1 37.3 36.1
School Parents ü ü 11.5 10.1 33.1 29.3
All Adults ü ü 20.1 20.0 25.3 18.7
All Ages ü ü ü ü 9.7 6.1 24.7 15.5
All Ages Wt Loss ü ü ü ü ü 5.3 6.1 14.7 15.1
50
Simulation-based Findings (1)
  • An inflection point in the growth of overweight
    and obesity prevalences probably occurred during
    the 1990s
  • Extrapolations assuming linear growth may
    therefore exaggerate future prevalences
  • The caloric imbalance relative to 1970 accounting
    for this growth has been only in the range of
    1-3 of daily caloric intake
  • Less than 50 kcal/dayper age, sex, and BMI
    category
  • Most of the overall observed increase in caloric
    intake (USDA CSFII 77-96 9 F, 13 M) has been
    the natural consequence of weight gain, not its
    cause
  • Both expenditure and intake naturally increase
    with greater weight

51
Reconciling the CSFII Data with Our Estimates of
Caloric Balance A Dynamic Hypothesis
Model Scope
Caloric balance
(up 1-2)
Mean BMI
(up 9-12)
52
Simulation-based Findings (2)
  • Current focus on interventions during childhood
    will have only small impact on overall adult
    obesity (6 relative to status quo)
  • Unless effectively linked to the rest of the
    population
  • Impacts on adult obesity of changing food and
    activity environments (by 2015) take decades to
    play out fully
  • Due to age-to-age carryover effect
  • Effective weight-loss programsif any existcould
    accelerate progress through subsidies for obese
    individuals
  • But the cost could be high (even if subsidies
    terminated by 2020)
  • And may be undermined by diet failure and
    recidivism

53
Conclusions
  • This model improves our understanding of
    population dynamics of weight change and supports
    pragmatic planning/evaluation
  • No other analytical model plays out effects of
    changes in caloric balance on BMI prevalences
    over the life-course
  • Traces plausible impacts of population-level and
    individual-level interventions
  • And addresses questions of whom to target, by how
    much, and by when
  • But it has limitationssome addressable, some due
    to lack of data
  • Does not indicate exact nature of interventions
  • Does not address cost-effectiveness of
    interventions, nor political reinforcement and
    resistance
  • Does not address racial/ethnic sub-groups
  • Does not trace individual life histories
    (compartmental structure)
  • Assumes habits determined by current environment,
    not by childhood learning
  • Assumes no irreversible metabolic changes
    sustained as a result of childhood
    overweight/obesity
About PowerShow.com