1. Crafting Integrated Strategies to Prevent and Manage Chronic Disease Using System Dynamics Chronic Disease Academy March 25, 2009 Seattle, WA

2. Presenters Phil Huang –Medical Director for City of Austin Department of Health and Human Services, formerly Chronic Disease Director for TX Patty Mabry –Office of Behavioral and Social Sciences Research, National Institutes of Health Bobby Milstein –Coordinator, Syndemics Prevention Network, Centers for Disease Control and Prevention Diane Orenstein –Technical Lead, Division for Heart Disease and Stroke Prevention, Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention Kris Wile –Sustainability Institute, System Dynamics Facilitator and Modeler

3. Workshop Agenda Wednesday, March 25 - Why are we here? - Introduction to System Analysis in Public Health - Introduction to the CV/Chronic Disease Risk Model - Testing Strategies for Reducing Preventable CVD and Chronic Disease Costs Demonstration Self-guided Exploration of Results Lunch - Building an Integrated Chronic Disease Strategy - Conclusions from model - Dialogue: Importance of Context - Lessons Learned - Dialogue: Future Opportunities Adjourn

4. Office of Behavior and Social Science’s Vision at NIH To mobilize the biomedical, behavioral, and social science research communities as partners in interdisciplinary research to solve the most pressing health challenges faced by our society. Programmatic Directions to Achieve the Vision: –Transdiciplinary science –“Next generation”, basic science –Problem-based, outcomes oriented strengthen the science of dissemination –Systems - thinking for population impact The Importance of Partnership for OBSSR

5. Adapted from Glass, McAtee (2006). Soc. Sci. Medicine, 62: 1650-1671 Health as a continuum between biological, behavioral and social factors across the lifespan and across generations

6. Simulation Modeling and Experimentation Pandemic flu Tobacco use Obesity, Diabetes Health inequalities “Non-health factors” Chronic disease Health care delivery Stress, mental illness, worksites, policy………. Understanding the “Whole” System

7. 200020012002200320042005 200620072008 Selected Examples from CDC’s Growing Portfolio of Simulation Studies for Health System Change SD Identified as a Promising Methodology for Health System Change Ventures Upstream- Downstream Dynamics Neighborhood Transformation Game National Health Economics & Reform Health Protection Game Overall Health Protection Enterprise Diabetes Action Labs Obesity Over the Lifecourse Fetal & Infant Health Syndemics Modeling* Cardiovascular Health in Context Selected Health Priority Areas

8. Questions Addressed by System Dynamics Modeling Exploring Strategies to Redirect the Course of Change Prevalence of Diagnosed Diabetes, US 0 10 20 30 40 19801990200020102020203020402050 Million people Historical Data Markov Model Constants Incidence rates (%/yr) Death rates (%/yr) Diagnosed fractions (Based on year 2000 data, per demographic segment) Honeycutt A, Boyle J, Broglio K, Thompson T, Hoerger T, Geiss L, Narayan K. A dynamic markov model for forecasting diabetes prevalence in the United States through 2050. Health Care Management Science 2003;6:155-164. Jones AP, Homer JB, Murphy DL, Essien JDK, Milstein B, Seville DA. Understanding diabetes population dynamics through simulation modeling and experimentation. American Journal of Public Health 2006;96(3):488-494. Why? Where? How? Who? What? Markov Forecasting Model Simulation Experiments in Action Labs

9. Time Series Models Describe trends Multivariate Stat Models Identify historical trend drivers and correlates Patterns Structure Events Increasing: Depth of causal theory Robustness for longer- term projection Value for developing policy insights Degrees of uncertainty Increasing: Depth of causal theory Robustness for longer- term projection Value for developing policy insights Degrees of uncertainty Dynamic Simulation Models Anticipate new trends, learn about policy consequences, and set justifiable goals Tools for Policy Planning & Evaluation

10. Different Modeling Approaches For Different Purposes Logic Models (flowcharts, maps or diagrams) System Dynamics (causal loop diagrams, stock-flow structures, simulation studies, action labs) Forecasting Models (regression models, Monte Carlo 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, better before worse) Test dynamic hypotheses through simulation studies Inspire action through visceral, game-based learning Make accurate forecasts of key variables Focus on precision of point predictions and confidence intervals

11. Brief Background on System Dynamics Modeling Compartmental models resting on a general theory of how systems change (or resist change) – often in ways we don’t expect –Developed for corporate policies in the 1950s, and applied to health policies since the 1970s –Concerned with understanding dynamic complexity Accumulation (stocks and flows) Feedback (balancing and reinforcing loops) –Used primarily to craft far-sighted, but empirically based, strategies Anticipate real-world delays and resistance Identify “high leverage” interventions –Modelers engage stakeholders through interactive workshops Forrester JW. Industrial Dynamics. Cambridge, MA: MIT Press; 1961. Sterman JD. Business Dynamics: Systems Thinking and Modeling for a Complex World. Boston, MA: Irwin/McGraw-Hill; 2000.

12. What is a System? What are Dynamics? System (Structure) = Stocks + Flows + Feedback Loops +… Stocks are accumulations of flows (of population, resources, changing goals, perceptions, etc.) Feedback loops link accumulations back to decisions that alter the flows: only 2 types (goal-seeking, self-reinforcing) Delays complicate things further As do non-linearities (need for critical mass, saturation effects) Dynamics = Behavior over time Patterns in time series data (growth, fluctuation, etc.) Visible relationships of two or more variables (move together, move opposite, lead-lag, etc.)

13. An (Inter) Active Form of Policy Planning/Evaluation System Dynamics is a methodology to… Map the salient forces that contribute to a persistent problem; Convert the map into a computer simulation model, integrating the best information and insight available; Compare results from simulated “What If…” experiments to identify intervention policies that might plausibly alleviate the problem; Conduct sensitivity analyses to assess areas of uncertainty in the model and guide future research; Convene diverse stakeholders to participate in model-supported “Action Labs,” which allow participants to discover for themselves the likely consequences of alternative policy scenarios

14. Simulations for Learning in Dynamic Systems Morecroft JDW, Sterman J. Modeling for learning organizations. Portland, OR: Productivity Press, 2000. Sterman JD. Business dynamics: systems thinking and modeling for a complex world. Boston, MA: Irwin McGraw-Hill, 2000. Multi-stakeholder Dialogue Dynamic Hypothesis (Causal Structure)Plausible Futures (Policy Experiments) Obese fraction of Adults (Ages 20-74) 0% 10% 20% 30% 40% 50% 197019801990200020102020203020402050 Fraction of popn 20-74

15. Getting Oriented Introduction –Name, Organization, What you do –What are you hoping to get out of today? Then talk with others at your tables: –What are the largest strategic issues you see in chronic disease? After 10 minutes, we’ll return to large group to share highlights –Biggest strategic challenges?

16. CDC Diabetes System Modeling Project Discovering Dynamics Through State-based Action Labs & Models Jones AP, Homer JB, Murphy DL, Essien JDK, Milstein B, Seville DA. Understanding diabetes population dynamics through simulation modeling and experimentation. American Journal of Public Health 2006;96(3):488-494.

17. Inflow Volume Outflow Developing Burden of Diabetes Total Prevalence (people with diabetes) Unhealthy Days (per person with diabetes) Costs (per person with diabetes) People with Diagnosed Diabetes Diagnosis Deaths a b People with Undiagnosed PreDiabetes Developing Diabetes Onset c d People with Normal Blood Sugar Levels PreDiabetes Onset Recovering from PreDiabetes e Diabetes Management Diabetes Diagnosis Obesity in the General Population PreDiabetes Detection & Management People with Undiagnosed Diabetes Deaths Diabetes Model: Diabetes Burden is Driven by Population Flows

18. Diabetes Burden is Driven by Population Flows Inflow Volume Outflow Developing Burden of Diabetes Total Prevalence (people with diabetes) Unhealthy Days (per person with diabetes) Costs (per person with diabetes) People with Diagnosed Diabetes Diagnosis Deaths a b People with Undiagnosed PreDiabetes Developing Diabetes Onset c d People with Normal Blood Sugar Levels PreDiabetes Onset Recovering from PreDiabetes e Diabetes Management Diabetes Diagnosis Obesity in the General Population PreDiabetes Detection & Management People with Undiagnosed Diabetes Deaths Standard boundary This larger view takes us beyond standard epidemiological models and most intervention programs

19. Diabetes System Dynamics Modeling Project Confirming Fit to Historical Trends (2 examples out of 10) Diagnosed Diabetes % of AdultsObese % of Adults 0% 10% 20% 30% 40% 1980198519901995200020052010 Obese % of adults Data (NHANES) Simulated 0% 2% 4% 6% 8% 1980198519901995200020052010 Diagnosed diabetes % of adults Data (NHIS) Simulated

20. The growth of diabetes prevalence since 1980 has been driven by growth in obesity prevalence Obese Fraction and Diabetes per Thousand 130 0.7 85 0.35 40 0 19801990200020102020203020402050 Time (Year) Diabetes Prevalence Obesity Prevalence Risk multiplier on diabetes onset from obesity = 2.6

21. Prevalence=92 AND RISING Baseline Scenario: Obesity to increase little after 2006, diabetes keeps growing robustly for another 20-25 years Obese Fraction and Diabetes per Thousand 130 0.7 85 0.35 40 0 19801990200020102020203020402050 Time (Year) Diabetes Prevalence Obesity Prevalence Diabetes prevalence keeps growing after obesity stops WHY? With high (even if flat) onset, prevalence tub keeps filling until deaths (4-5%/yr)=onset Onset=6.3 per thou Estimated 2006 values Death=3.8 per thou Risk multiplier on diabetes onset from obesity = 2.6

22. Unhealthy days impact of prevalence growth, as affected by diabetes management: Past and one possible future Unhealthy Days per Thou and Frac Managed Obese Fraction and Diabetes per Thousand 130 0.7 85 0.35 40 0 19801990200020102020203020402050 Time (Year) Diabetes Prevalence Obesity Prevalence 500 0.65 250 0 19801990200020102020203020402050 375 0.325 Unhealthy Days from Diabetes Managed fraction Diabetes prevalence keeps growing after obesity stops If disease management gains end, the burden grows Reduction in unhealthy days per complicated case if conventionally managed: 33%; if intensively managed: 67%

23. A Sequence of What-if Simulations People with Diabetes per Thousand Adults 150 125 100 75 50 19801990200020102020203020402050 Monthly Unhealthy Days from Diabetes per Thou 500 450 400 350 300 250 19801990200020102020203020402050 Base Start with the base case or “status quo”: no improvements in diabetes management or prediabetes management after 2006

24. What if there were further Increases in Diabetes Management? People with Diabetes per Thousand Adults 150 125 100 75 50 19801990200020102020203020402050 Monthly Unhealthy Days from Diabetes per Thou 500 450 400 350 300 250 19801990200020102020203020402050 Base Diab mgt Base More people living with diabetes Keeping the burden at bay for nine years longer Diab mgt Increase fraction of diagnosed diabetes getting managed from 58% to 80% by 2015. (No change in the mix of conventional and intensive.) What do you think will happen? Diabetes mgmt does nothing to slow the growth of prevalence—in fact, it increases it. As soon as diabetes mgmt stops improving, unhealthy days start to grow as fast as prevalence.

25. What if there was a huge push for Prediabetes Management? People with Diabetes per Thousand Adults 150 125 100 75 50 19801990200020102020203020402050 Monthly Unhealthy Days from Diabetes per Thou 500 450 400 350 300 250 19801990200020102020203020402050 Base PreD mgmt Base PreD mgmt The improvement is relatively modest—the growth is not stopped Increase fraction of prediabetics getting managed from 6% to 32% by 2015. (Half of those under intensive mgmt by 2015.) No increase in diabetes mgmt. What do you think will happen? Diabetes onset rate reduced 12% relative to base run. Not nearly enough to offset the excess onset due to high obesity. By 2050, diabetes prevalence reduced only 9% relative to base run.

26. Diabetes Model: What if Obesity is Reduced? Two Scenarios Obese Fraction of Adult Population 0.4 0.3 0.2 0.1 0 19801990200020102020203020402050 Base Obesity 25% Obesity 18% What if it were possible—in addition to the prediabetes mgmt intervention - to gradually lower the fraction obese from 34% (2006) to the 1994 value of 25% by 2030? Or, to the 1984 value of 18%?

27. Diabetes: What if we Managed Prediabetes AND Reduced Obesity? The more you reduce obesity, the sooner you stop the growth in diabetes—and the more you bring it down … Same with the burden of diabetes People with Diabetes per Thousand Adults 150 125 100 75 50 19801990200020102020203020402050 Monthly Unhealthy Days from Diabetes per Thou 500 450 400 350 300 250 19801990200020102020203020402050 Base PreD mgmt PreD & Ob 25% PreD & Ob 18% Base PreD mgmt PreD & Ob 18% PreD & Ob 25% What do you think will happen if, in addition to PreD mgmt, obesity is reduced moderately by 2030? What if it is reduced even more? Why is obesity reduction so powerful? Mainly because of its strong effect on onset rate among prediabetics; but, also, because it reduces PreD prevalence itself. However, achieving significant obesity reduction takes a long time.

28. What if Intervened Effectively Upstream AND Downstream People with Diabetes per Thousand Adults 150 125 100 75 50 19801990200020102020203020402050 Monthly Unhealthy Days from Diabetes per Thou 500 450 400 350 300 250 19801990200020102020203020402050 Base PreD mgmt Base PreD & Ob 25% Pred & Ob 25% All 3 -- PreD & Ob 25% & Diab mgmt All 3 With a combination of effective upstream and downstream interventions we could hold the burden of diabetes nearly flat through 2050! With pure upstream intervention, burden still grows for many years before turning around. What do you think will happen if we add the prior diabetes mgmt intervention on top of the PreD+Ob25 one? Downstream improvement acts quickly against burden but cannot continue forever. Significant upstream gains are thus essential but will likely take 15+ years to achieve. A flat-burden future is possible but requires simultaneous action on both fronts.

29. CDC Obesity Dynamics Modeling Project Contributors 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 System Dynamics Consultants Jack Homer Gary Hirsch Time Series Analysts Danika Parchment Cynthia Ogden Margaret Carroll Hatice Zahran Project Coordinator Bobby Milstein Workshop Participants Atlanta, GA: May 17-18 (N=47) Lansing, MI: July 26-27 (N=55) Homer J, Milstein B, Dietz W, Buchner D, Majestic D. Obesity population dynamics: exploring historical growth and plausible futures in the U.S. 24th International Conference of the System Dynamics Society; Nijmegen, The Netherlands; July 26, 2006. Cover of "The Economist", Dec. 13-19, 2003 Cover of "The Economist", Dec. 13-19, 2003.

30. Focusing on Life-Course Dynamics Explore likely consequences of possible interventions affecting caloric balance (intake less expenditure) –How much impact on obesity prevalence? –How long will it take to see? –Should we target particular subpopulations? (age, sex, weight category; lack data for race, ethnicity) Consider interventions broadly but leave details (composition, coverage, efficacy, cost) outside model boundary for now –Available data inadequate –Would require a separate research effort to estimate these details –Not addressing feedback loops of reinforcement and resistance –Not addressing cost-effectiveness

31. Obesity Dynamics Over the Decades Dynamic Population Weight Framework Dynamic Population Weight Framework Population by Age (0-99) and Sex Flow-rates between BMI categories Overweight and obesity prevalence Birth Immigration Death Caloric Balance Yearly aging Not Overweight Moderately Overweight Moderately Obese Severely Obese Trends and Planned Interventions Changes in the Physical and Social Environment Weight Loss/Maintenance Services for Individuals Data source: National Center for Health Statistics, CDC: National Health Examination Survey (NHES) 1960- 1970, National Health and Nutrition Examination Survey (NHANES) 1971-2002. Homer J, Milstein B, Dietz W, et al. Obesity population dynamics: exploring historical growth and plausible futures in the U.S. Proc. 24th Int’l System Dynamics Conference; Nijmegen, The Netherlands; July 2006.

32. Alternative Futures for Adult Obesity Obese fraction of Adults (Ages 20-74) 0% 10% 20% 30% 40% 50% 197019801990200020102020203020402050 Fraction of popn 20-74 BaseSchoolYouthAllYouth School+ParentsAllAdultsAllAges AllAges+WtLoss

33. Results of Simulated Interventions Environmental change approach (reduce caloric balances to their 1970 values by 2015 for selected age ranges) Youth interventions have only small impact on overall adult obesity (assuming adult habits determined by adult environments—not by childhood 1 ) Slow decline in overall adult obesity, even when program covers all ages Targeted weight loss approach (obese lose 4 lbs per year, program terminated 2020) Such a program could accelerate progress and “buy time” for environmental change (but first, need to find a cost-effective program with lasting benefits—minimal relapse) Need to assure caloric balance throughout all ages, particularly adulthood. Contrast today’s narrow national focus on school-age youth. Also need research on extent to which adult habits are determined by childhood. 2 Need to assure caloric balance throughout all ages, particularly adulthood. Contrast today’s narrow national focus on school-age youth. Also need research on extent to which adult habits are determined by childhood. 2 1. Christakis and Fowler. NEJM 357, 2007. 2. Bar-Or O., PCPFS Research Digest Series 2, No. 4, 1995.

34. Simulating the Dynamics of Cardiovascular Health and Related Risk Factors Work in Progress This work was funded by the CDC’s Division for Heart Disease and Stroke Prevention and by the National Institutes of Health’s Office of Behavioral and Social Science Research. The work was done in collaboration with the Health and Human Services Department of Austin/Travis County, Texas, and with Integrated Care Collaboration of Central Texas. The external contractors are Sustainability Institute and RTI International. Homer J, Milstein B, Wile K, Pratibhu P, Farris R, Orenstein D. Modeling the local dynamics of cardiovascular health: risk factors, context, and capacity. Preventing Chronic Disease 2008;5(2). Available at http://www.cdc.gov/pcd/issues/2008/apr/07_0230.htm Homer J, Milstein B, Wile K, Trogdon J, Huang P, Labarthe D, Orenstein D. Simulating and evaluating local interventions to improve cardiovascular health. In submission to Preventing Chronic Disease.

35. Cardiovascular Disease and Risks Remain Among the Leading Causes of Death United StatesTexas 1. Heart Disease26.6%1. Heart Disease25.7% 2. Cancer22.8%2. Cancer21.9% 3. Stroke5.9%3. Stroke6.0% 4. Chronic Lower Respiratory Disease 5.3%4. Accidents5.5% 5. Accidents4.8% 5. Chronic Lower Respiratory Disease 5.1% 6. Diabetes3.1%6. Diabetes3.6% *US: CDC/National Center for Health Statistics, Vol. 56, No.10, April 2008; TX: TX Dept. of State Health Services Preliminary Vital Statistics Table 16 Fraction of total deaths in 2005*…

36. Reducing Disability & Risk of Recurrent CVD Detecting & Treating Acute CVD Events Controlling Increased CVD Risk Preserving Low CVD Risk From Healthy People 2010: 4 Levels of Prevention for Cardiovascular Diseases

37. Disability and Risk of CVD Recurrence Acute CVD Events Increased CVD Risk Low CVD Risk 4 levels of prevention correspond to 4 States of Cardiovascular Health:

38. NUTRITION, PHYSICAL ACTIVITY & STRESS Salt intake Saturated/Trans fat intake Fruit/Vegetable intake Net caloric intake Physical activity Chronic stress CVD RISK FACTOR PREVALENCE & CONTROL Hypertension High cholesterol Diabetes Obesity Smoking Secondhand smoke Air pollution exposure UTILIZATION OF SERVICES Behavioral change Social support Mental health Preventive health COSTS (CVD & NON-CVD) ATTRIBUTABLE TO RISK FACTORS LOCAL CONTEXT Eating & activity options Smoking policies Socioeconomic conditions Environmental policies Health care options Support service options Media and events Local capacity for leadership & organizing LOCAL ACTIONS ESTIMATED FIRST- TIME CVD EVENTS CHD (MI, Angina, Cardiac Arrest) Stroke Total CVD (CHD, Stroke, CHF, PAD) Preventing and Managing Risk Factors for CVD Disability and Risk of CVD Recurrence Acute CVD Events Increased CVD Risk Low CVD Risk

39. NUTRITION, PHYSICAL ACTIVITY & STRESS Salt intake Saturated/Trans fat intake Fruit/Vegetable intake Net caloric intake Physical activity Chronic stress CVD RISK FACTOR PREVALENCE & CONTROL Hypertension High cholesterol Diabetes Obesity Smoking Secondhand smoke Air pollution exposure UTILIZATION OF SERVICES Behavioral change Social support Mental health Preventive health COSTS (CVD & NON-CVD) ATTRIBUTABLE TO RISK FACTORS LOCAL CONTEXT Eating & activity options Smoking policies Socioeconomic conditions Environmental policies Health care options Support service options Media and events Local capacity for leadership & organizing LOCAL ACTIONS ESTIMATED FIRST- TIME CVD EVENTS CHD (MI, Angina, Cardiac Arrest) Stroke Total CVD (CHD, Stroke, CHF, PAD) Interventions Through Local Context Homer J, Milstein B, Wile K, Pratibhu P, Farris R, Orenstein D. Modeling the local dynamics of cardiovascular health: risk factors, context, and capacity. Preventing Chronic Disease (in press).

40. Purpose of the Cardiovascular Risk Model How do local conditions affect multiple risk factors for CVD, and how do those risks affect population health status and costs over time? How do different local interventions affect cardiovascular health and related expenditures in the short- and long-term? How might local health leaders better balance their policy efforts given limited resources? The CDC has partnered with the Austin (Travis County), Texas, Dept. of Health and Human Services. The model is calibrated to represent the overall US, but is informed by the experience and data of the Austin team, which has been supported by the CDC’s “STEPS” program since 2004. The CDC has partnered with the Austin (Travis County), Texas, Dept. of Health and Human Services. The model is calibrated to represent the overall US, but is informed by the experience and data of the Austin team, which has been supported by the CDC’s “STEPS” program since 2004. Homer J, Milstein B, Wile K, Pratibhu P, Farris R, Orenstein D. Modeling the local dynamics of cardiovascular health: risk factors, context, and capacity. Preventing Chronic Disease 2008;5(2). Available at http://www.cdc.gov/pcd/issues/2008/apr/07_0230.htm

41. Direct Risk Factors

42. Indirect Risk Factors

43. Tobacco and Air Quality Interventions

44. Air Quality Interventions

45. Health Care Interventions

46. Interventions Affecting Stress

47. Healthy Diet Interventions

48. Physical Activity & Weight Loss Interventions

49. Adding Up the Costs

50. Cardiovascular event costs Medical costs (ER, inpatient, rehab)—for non-fatal & fatal events Productivity (morbidity) losses* from non-fatal events Productivity (premature mortality) losses* from fatal events Non-cardiovascular complications of risk factors Hospital costs due to non-CV complications of diabetes (e.g., kidneys, eyes, feet), high BP, & smoking Productivity (morbidity) losses* from non-fatal complications of diabetes, high BP, smoking, & obesity Productivity (mortality) losses* from fatal complications of smoking (e.g., cancer, COPD), diabetes, high BP, & obesity Costs of managing risk factors Medications & visits for diabetes, high BP, high cholesterol—by level of care (high quality = 2 – 2.5x cost of mediocre care) Other services: Mental health services, Weight loss services, Smoking quit services & products Human capital approach based on: Haddix, Teutsch, Corso, Prevention Effectiveness, 2003 (2nd ed, Tables 1.1b and 1.1c).

51. Relative size of included Complication Costs CV Events Non-CV Complications of Diabetes Non-CV Complications of Hypertension Non-CV Complications of Smoking Non-CV Complications of Obesity Direct medical costs of complications ++++ * +*+* +*+* 0*0* Indirect productivity losses: disability ++++ * +*+* +*+* +*+* Indirect productivity losses: premature death ++ + + * Non-CV hospitalization costs & lost workdays estimated from MEPS 2000-03 linked with NHIS. The regression analysis controlled for demographics, CVD, and unrelated diseases (e.g., HIV).

52. Data Sources for Modeling CVD Risk Census –Population, deaths, births, net immigration, health coverage AHA & NIH statistical reports –Cardiovascular events, deaths, and prevalence (CHD, stroke, CHF, PAD) National Health and Nutrition Examination Survey (NHANES) –Risk factor prevalences by age (18-29, 30-64, 65+) and sex (M, F) –Chronic disorder diagnosis and control (hypertension, high cholesterol, diabetes) Behavioral Risk Factor Surveillance System (BRFSS) –Diet & physical activity –Primary care utilization –Lack of needed emotional/social support  Psychosocial stress Medical Examination Panel (MEPS) / National Health Interview (NHIS) –Medical and productivity costs attributable to smoking, obesity, and chronic disorders Research literature –CVD risk calculator, and relative risks from SHS, air pollution, obesity, and inactivity –Medical and productivity costs of cardiovascular events Questionnaires for CDC and Austin teams (expert judgment) –Potential effects of social & services marketing on utilization behavior –Effects of behavioral services on smoking, weight loss, stress reduction –Relative risks of stress for high BP, high cholesterol, smoking, and obesity

53. Calculating First-Time CV Events & Deaths Based on well-established Framingham approach for calculating probability of first-time events & deaths in individuals CVD = CHD (MI, angina, cardiac arrest) + Stroke/TIA + CHF + PAD Modifies individual-level risk calculator for use with populations Uses prevalences of uncontrolled chronic disorders by sex/age group Introduces secondhand smoke and pollution as additional risk factors Combines risks multiplicatively to account for overlapping conditions Adjustment exponents reproduce synergies seen in individual-level calculator Adjustment multipliers reproduce AHA event and death frequencies for 2003 - Anderson et al, Am Heart J 1991 (based on Framingham MA population N=5573, 1968-1987) - Homer “Risk calculation in the CVD model” project document, June 19, 2007 - NHANES 1988-94 & 1999-04 - AHA Heart Disease and Stroke Statistics – 2006 Update

54. Interactive Model Guide Details key assumptions and sources of evidence for each relationship in the model On the CD ROM for participant today –Called “CVD interactive guide v8m.ppt” –Functions in slide show mode Afterward, we will have an opportunity for remote Q&A with Jack Homer, Lead Modeler, and Justin Trogdon, Cost data expert.

55. Where are your efforts to manage chronic disease?

56. Individual Strategy Fly-by Exercise Your goal is to reduce CV deaths and reduce total risk factor consequence costs … but you have limited resources. Pick 6 interventions for your strategy Record your selected interventions on your sheet under Section #1 Questions? You have 10 minutes.

57. A Base Case Scenario for Comparison Assumptions for Input Time Series through 2040 A plausible and straightforward scenario –Assume no further changes in contextual factors affecting risk factor prevalences –Any changes in prevalences after 2004 are due to “bathtub” adjustment process and population aging –Provides an easily-understood basis for comparisons Prior to 2004, model reflects declining … –Fraction workplaces allowing smoking (1990-2003) –Air pollution (1990-2001) –Youth smoking (rise 1991-99, decline 1999-2003) –CV event fatality (1990-2003) Total RF Complication Costs per Capita 2,000 1,000 0 199020002010202020302040 Complication Costs per 1000 if all risk factors = 0 Also note: Cost minimum if all proximal risk factor prevalences were zero. Consequence costs would decrease 80% CV death rate would be 60% below the base case. 3,000 No Further Changes in Drivers

58. Base run behaviors CVD & Risk Factor Complication Costs and CVD Mortality 0.6 0.3 0 199020002010202020302040 Smoking Prevalence Air Pollution PM2.5 Diabetes Prevalence High BP Prevalence High Cholesterol Prevalence CV Risk Factor Prevalences 30 15 0 Obese Adults Newly obese adults Becoming non-obese or dying 2040 0 0.4 % Obese 1990 Result: Past trends level off after 2004, after which results reflect only slow “bathtub” adjustments in risk factors Increasing obesity, high BP, and diabetes Decreasing smoking and air pollution Increases in risk factors and population aging lead to eventual rebound in deaths (Air pollution only) 199020002010202020302040 4 3 2 1 0 Deaths from CVD per 1000 if all risk factors = 0 Deaths from CVD per 1000 Complication Costs per 1000 Complication Costs per 1000 if all risk factors = 0 3,000 2,250 1,500 750 0

59. Base case behavior for 1990-2040 1 0 Use of Primary Care Services 0.3 0 Stress Prevalence 0.8 0 Poor Diet Fraction 0.8 0 Inadequate Physical Activity 4 0 0.3 0 Smoking Prevalence 0.4 0 Obesity Prevalence 0.6 0 Secondhand Smoke Exposure 0.6 0 Diabetes High BP High cholesterol 30 0 Particulate Air Pollution PM2.5 3,000 0 CVD & Risk factor costs per capita Uncontrolled CVD Deaths per 1000 Prevalences mcg per m 3 Age 65+ fraction of the population CV event fatality multiplier 0.3 0 1.5 0

60. Area of effect and type of intervention –Increasing access –Marketing of services –Social marketing –Taxes and/or sales restriction –Others Intervention Options

61. Tracing interventions through the system- Increasing access to physical activity options

62. Interpreting Cost Results Complication costs are for CV and non-CV related complications, both direct and indirect Management costs include –Annual costs for services provided –Medication costs When these costs are less than baseline, the difference is the per capita health cost savings per year – the maximum economically justifiable spending for the intervention Complication & Management Costs per Capita 3,000 2,000 199020002010202020302040 * Average annual savings of *$ 49 per capita from interventions to increase access to physical activity options from 2010 - 2040. Base Case Increased Access to Physical Activity options

63. Develop and Test Your Team Strategies Form groups of 4 or 5 people Same goals – Reduce CV deaths and total risk factor consequence costs with limited resources. Choose 6 interventions for your strategy Prepare a flipchart to present your strategy. You have 15 minutes. We will test some team strategies with the simulator.

64. Interactive Results Exploration Purpose: To develop conclusions about which strategies are most effective to achieve our goals 1.Work in groups of 2-3 with a laptop. 2.As you explore, fill out Section #2 of worksheet. –Which interventions were more or less powerful than you expected? –Your conclusions? Ideas? Please note your questions so we can explore them in the larger group. You have 25 minutes.

65. Break for Lunch We will begin again promptly at 1:00 pm. This afternoon, we will: –Revise and test team strategies. –Share conclusions from this systemic analysis. –Explore opportunities

66. Revising Team Strategies 1.Teams choose their goals. What are you trying to achieve? What is your timeframe? 2.Put together a strategy, choosing six different interventions that you think will work. 3.Record on flipchart. 4.We will try them out with the simulator. 5.You have 10 minutes.

67. Conclusions: Comparing intervention groups Care Primary Care Quality = 75% PC Marketing = 100% PC Access = 100% Lifestyle Physical Activity Access = 100% Physical Activity Social Marketing = 100% Access to Healthy Nutrition = 100% Healthy Nutrition Social Marketing = 100% Stress Multiplier = ½ Air Tobacco Tax = 100% Marketing Against Smoking = 100% Air Pollution Multiplier = ½ Smoking Bans = 100%

68. Comparing Care, Air & Lifestyle Interventions Care provides –quick and sustained reduction in CV events, –but little cost savings. Air provides –quick and growing reduction in CV events, –and major cost savings. Lifestyle provides –Growing CV event reductions over time, but little immediately –Substantially increasing cost savings over time Deaths from CVD per 1000 4 2 0 199020002010202020302040 Base Case Care Care + Air Care + Air + Lifestyle If all risk factors = 0 Complication & Mgmt Costs per Capita 3,000 0 199020002010202020302040 Base Case Care Care + Air Care + Air + Lifestyle If all risk factors = 0

69. Cost Conclusions AIR – Smoking and air quality interventions can save lives quickly and can justify intervention spending up to $300 per capita for 30 years ($355 in ET). CARE – Improving utilization and quality of primary care services can save lives quickly, but should not be expected to save much on total costs. Justified intervention spending could be up to $25 per capita for 30 years ($35 in ET). LIFESTYLE – Improving nutrition and physical activity, and reducing sources of stress take longer to affect CV events though obesity and chronic conditions. However their contribution grows over time and intervention spending of up to $100 per capita could be justified ($177 in ET).

70. Comparing E. Travis to US: More Effective Individual Interventions After 10 years (2015)After 35 years (2040) CVD Death Rate Compl + Mgmt Costs CVD Death Rate Compl + Mgmt Costs ETUSETUSETUSETUS Social Marketing Against Smoking 44114311 Quality of Primary Care 113 11 Tobacco Tax and Sales Restrictions 522 22 Air Pollution 324*352 4 Access to Primary Care 234424 Access to Physical Activity 5 2433 Access to Healthy Diet 44 Stress Reduction 4 5 *Duplicates ranks indicate ties.

71. Overall Conclusions CV death rate has declined due to improvements in acute care, and also reductions in smoking, second hand smoke, and air pollution. Risk factor consequence costs have decreased as a consequence, but also because of reductions in smoking related deaths. Smoking will probably continue to decline in growing elderly population, helping to lower costs. Of 19 interventions, at least 15 have the potential to reduce CVD events without increasing costs.

72. Revise Individual Strategies With this additional information, would you adjust your individual strategy? Why? Please record selected interventions on your sheet under Section #3 You have 10 minutes.

73. Integrating your contextual knowledge for integrated public health policy 1.Are there gaps in specific policy areas? Where? 2.Where are the opportunities? 3.What new partnerships can be identified now?

74. How we can make it happen 1.What partnerships or processes need to be created or strengthened? 2.Make a personal commitment To making those connections and to collaborating Indicate your commitment in Section 5 of your worksheet

75. Model Boundaries / Limitations More contextual information must come into play. Decision support tool to inform multi-stakeholder dialogue. Local experts provide crucial link to relevant data and implementation.

76. Societal Dialogue Incorporates Model Omissions Other chronic disease endpoints Downstream interventions and costs Local implementation opportunities Local implementation strengths and success Political will What measures of improvement ought to be included? What else is missing? What would be helpful to you? SYSTEM DYNAMICS MODEL STRATEGIC DIALOGUE Implementation actions and costs Health inequities Local leadership capacity Ability to engage all stakeholders Borderline conditions

77. What we have learned The simulator and surrounding dialogue can be used to: –Create alignment among stakeholders –Initiate systemic thinking, increasing leadership capacity –Spur people to action –Identify opportunities and build commitment to address them –Inform the development of business cases for investment in interventions

78. Sample Austin Implementation Worksheet Intervention Area SchoolsCommunityWorksitesHealthcare Physical Activity Nutrition Tobacco CVD Diabetes Cancer

79. Integrative Modeling for Strategy Building Integrating large and varied sources of evidence in simulations is fundamentally useful. Complex problems, those with many causal pathways and significant time delays, are suitable. All models are limited, so they are useful as decision-support tools. Local context affects appropriate strategy. Translucent box – making assumptions visible can develop trust and leadership capacity. Don’t forget the “ask”. Give people an opportunity to take action.

80. What opportunities do you see for integrated policy making? Other chronic disease endpoints Implementation actions and costs Downstream interventions and costs Health inequities Local implementation opportunities Local implementation strengths and success Local leadership capacity Political will Ability to engage all stakeholders We plan to extend this model –Borderline conditions, ex- smokers –Downstream interventions and costs Investigate transferability of this model to other locales Tools allowing wider dissemination What do need for this to be useful to you? Needs for other systemic analyses? SYSTEM DYNAMICS MODEL STRATEGIC DIALOGUE Borderline conditions

81. OBSSR at NIH Vision: To mobilize the biomedical, behavioral, and social science research communities as partners in interdisciplinary research to solve the most pressing health challenges faced by our society. 27 NIH Institutes and Centers and the extramural community. Programmatic Directions to Achieve the Vision: –Trans-/inter-disciplinary science –“Next generation”, basic science –Problem-based, outcomes oriented strengthen the science of dissemination –Systems science for population impact

82. What is the challenge? Other approaches alone have not solved intractable health problems Health problems are embedded in dynamically complex systems Policies, programs, interventions have limited resources and involve trade offs Could try “kitchen sink” approach, but resources are limited Could try a “thought experiment” but the human mind cannot execute beyond simple

83. Systems Science Activities at NIH 2007 Symposia Series on Systems Science & Health Institute on Systems Science and Health (annually) BSSR-Systems Science listserv - send email to mabryp@od.nih.gov mabryp@od.nih.gov CDC SD Modeling with OBSSR and NHLBI

84. Examples of NIH Modeling Initiatives Cancer Intervention and Surveillance Modeling Network (CISNET): http://cisnet.cancer.gov/about/ Interagency Modeling and Analysis Group (IMAG): http://www.imagwiki.org/mediawiki http://grants.nih.gov/grants/guide/pa-files/PAR-08-023.htm Models of Infectious Disease Agent Study (MIDAS): http://www.nigms.nih.gov/Initiatives/MIDAS NIH Guide To Grants And Contracts http://grants.nih.gov/grants/guide/index.html To Subscribe to the NIH Guide LISTSERV, send an e-mail to listserv@list.nih.gov with the following text in the message body (not the "Subject" line): subscribe NIHTOC-L your name

85. Active NIH FOA’s in Systems Science and BSSR PAR-08-224 Using Systems Science Methodologies to Protect and Improve Population Health (R21). Expires Sept 2011. 3 receipt dates per year. Contact Patty Mabry, OBSSR. PAR-08-212, -213, -214 Methodology and Measurement in the Behavioral and Social Sciences (R01, R21, R03). Expires September 2011. 3 receipt dates per year. Contact Deb Olster, OBSSR. RFA-07-079, -080 Behavioral and Social Science Research on Understanding and Reducing Health Disparities (R01, R21) Expires September 2009. One receipt date per year Sept. Contact: Ron Abeles, OBSSR. PAR-08-023 Predictive Multiscale Models of the Physiome in Health and Disease (R01). Expires September 2010. 3 receipt dates per year. Contact: Grace Peng, NIBIB.

86. Grant Funded Systems Science and BSSR at NIH Joshua Epstein, Director’s Pioneer Award, NIGMS, OBSSR, 2008. Project Title: Behavioral Epidemiology: Applications of Agent-Based Modeling to Infectious Disease. David Lounsbury, R03, NIDA, 2008. Project Title: Dynamics Modeling as a Tool for Disseminating the PHS Tobacco Treatment Guideline David T. Levy, U01, NCI, 2002-2010. CISNET. Project Title: A Simulation of Tobacco Policy, Smoking and Lung Cancer. Linda Collins & Daniel Rivera, R21, 2007-2010. NIH Roadmap. Dynamical System /Related Engineering Approach /Improving Behavioral Intervention Daniel Rivera, K25, NIDA, OBSSR. Control Engineering Approaches to Adaptive Interventions in Drug Abuse Prevention. PAR-08-224 – Awards pending. RFA-HD-08-023 (R01), Innovative Computational and Statistical Methodologies for the Design and Analysis of Multilevel Studies on Childhood Obesity (R01). Awards pending.

87. For more information: Patty Mabry, Ph.D. mabryp@od.nih.gov Office of Behavioral and Social Sciences Research (OBSSR) National Institutes of Health http://obssr.od.nih.gov

88. Check-out and evaluation Please fill out Sections 4 and 5 of your worksheet. Do you have any thoughts you would like to offer? Questions? Reactions? Plans? Feedback?

89. Chronic Disease Leaders THANK YOU! Small favors: Please leave your Strategy Worksheets with us. Take a CD-ROM for your own use. We encourage you to share.