How to stem the US opioid epidemic and reduce its damage?

How to allocate state funds for preventing cancer, cardiovascular disease, and pulmonary disease?

How to improve population health and well-being in a metropolitan area?

How to reduce cardiovascular disease risk in a developing country?

How to accelerate the reduction of malnutrition-related stunting in the developing world?

How to reform local health systems to improve health and reduce costs, with sustainable financing?

How to make Veterans Administration treatment of Hepatitis C more cost-effective?

How to protect patients and health care providers in the face of Medicare payment rate cuts?

How to change U.S. health policy so that it simultaneously reduces costs, illness, and disparities?

How to encourage hospital adoption of programs to prevent healthcare-associated infections?

How to reduce adverse health outcomes and health disparities in a socially diverse urban center?

How to allocate Veterans Administration funds for the prevention and treatment of strokes?

How to reduce the burden of AIDS in a developing country through community-based services?

How to reduce the future health burden of cardiovascular risk factors in the U.S.?

How to improve the capacity of a hospital to handle a disaster-related surge in demand?

How to improve educational attainment in a developing country?

Published Projects in Detail


Local Health System Reform Strategy

Client:  Fannie E. Rippel Foundation (New Jersey)
Publication:  Hirsch G, Homer J, Milstein B, and 5 others.  ReThink Health Dynamics: Understanding and Influencing Local Health System Change.  Proceedings of the 30th International System Dynamics Conference, St. Gallen, Switzerland, 2012.

Health system reform is a national priority in the U.S., but it is increasingly being pursued through a mosaic of local initiatives.  Concerned leaders in cities, towns, and regions across the country are working within their local health systems to achieve better health, better care, lower cost, and greater equity.  Such ambitious ventures are, however hard to plan, unwieldy to manage, and slow to spread.  Further progress could occur if diverse stakeholders were better able to play out intervention scenarios, weigh trade-offs, set aside schemes that are unlikely to succeed, and enact strategies that promise the most robust results.  Through the Rippel Foundation’s ReThink Health initiative and the ReThink Health Dynamics simulation model, local leaders are learning what it takes to spark and sustain system-wide improvements in their settings. 

Social Determinants of Health in a Diverse Urban Population

Client:  Wellesley Institute (Toronto)
Publication:  Mahamoud A, Roche B, Homer J.  Modeling the Social Determinants of Health and Simulating Short-Term and Long-Term Intervention Impacts for the City of Toronto, Canada.  Social Science and Medicine, published online 10 October 2012.  DOI: 10.1016/j.socscimed.2012.06.036.

Social determinants of health are important in shaping the health of urban populations in Canada. The low socioeconomic status of marginalized, disadvantaged, and precarious populations in urban settings has been linked to adverse health outcomes including chronic and infectious disease, negative health behaviors, barriers to accessing health care services, and overall mortality. Given the dynamic complexities and interrelationships surrounding the underlying drivers of population health outcomes and inequities, it is difficult to assess program and policy intervention tradeoffs, particularly when such interventions are studied with static models. To address this challenge, a simulation model was developed for the City of Toronto, Canada, utilizing system dynamics modeling methodology. The model simulates changes in health, social determinants, and disparities from 2006 and projects forward to 2046 under different assumptions. Most of the variables in the model are stratified by ethnicity, immigration status, and gender, and capture the characteristics of adults aged 25–64. Intervention areas include health care access, behavior, income, housing, and social cohesion. The model simulates alternative scenarios to help demonstrate the relative impact of different interventions on poor health outcomes such as chronic disease rates, disability rates, and mortality rate. It gives insight into how much, and how quickly, interventions can reduce mortality and morbidity. This will serve as a useful learning tool to allow diverse stakeholders and policy makers to ask “what if” questions and map effective policy directions for complex population health problems, and will enable communities to think about their health futures.

National Health System Reform Strategy

Client:  Centers for Disease Control and Prevention (CDC)

Milstein B, Homer J, Hirsch G.  Analyzing National Health Reform Strategies with a Dynamic Simulation Model. American Journal of Public Health, 100(5):811-819, 2010.

Milstein B, Homer J, Briss P, Burton D, Pechacek T.  Why Behavioral and Environmental Interventions are Needed to Improve Health at Lower Cost.  Health Affairs, 30(5), May 2011.  DOI: 10.1377/hlthaff.2010.1116.

Proposals to improve the US health system are commonly supported by models that have only a few variables and overlook certain processes that may delay, dilute, or defeat intervention effects. An evidence-based dynamic simulation model with a broad national scope was developed and used to analyze various policy proposals. The results suggest that expanding insurance coverage and improving health care quality would likely improve health status but would also raise costs and worsen health inequity, whereas a strategy that also strengthens primary care capacity and emphasizes health protection would improve health status, reduce inequities, and lower costs. A software interface allows diverse stakeholders to interact with the model through a policy simulation game called HealthBound.

National Chronic Disease Strategy

Client:  Centers for Disease Control and Prevention (CDC) and National Heart, Lung, and Blood Institute (NHLBI)

Recent Publications: 

Homer J, Wile K, Yarnoff B, Trogdon JG, Hirsch G, Cooper L, Soler R, Orenstein D. Using Simulation to Compare Established and Emerging Interventions to Reduce Cardiovascular Disease Risks in the United States. Preventing Chronic Disease, 11(E195):1-14, November 2014. Available at:  

Hirsch G, Homer J, Wile K, Trogdon JG, Orenstein D.  Using Simulation to Compare 4 Categories of Intervention for Reducing Cardiovascular Disease Risks. American Journal of Public Health, 104(7):1187-1195, 2014.

At least 70% of deaths among Americans each year are from chronic diseases, and their direct and indirect costs are more than 1 trillion dollars per year.  Governmental health agencies are in a position to promote strategies to prevent and manage chronic disease, but identifying the most effective and economical strategies is often difficult.  To help health agencies better plan and evaluate interventions, the CDC and the NHLBI funded the creation of the Prevention Impacts Simulation Model (PRISM).  PRISM is a relatively large system dynamics model that is used to simulate trajectories for health and cost outcomes for the entire U.S. population from 1990 to 2040, and has also been applied to represent other national and local populations.  Interventions are in several broad areas: medical care, smoking, nutrition and weight loss, physical activity, emotional distress, and particulate air pollution.  These interventions act through a range of channels such as access, price, promotion, and regulation.  The diseases and conditions modeled in detail include heart disease, stroke, diabetes, hypertension, high cholesterol, and obesity, and the model also accounts for cancers and respiratory diseases related to smoking, obesity, poor nutrition, and physical inactivity.  The model reports summary measures of mortality and years of life lost as well as the consequent medical and productivity costs of the chronic diseases and conditions modeled.  Local and federal health officials have used PRISM throughout its development, and its applications continue to grow in number and variety.

Obesity Population Dynamics

Client:  Centers for Disease Control and Prevention (CDC)
Publication:  Homer J, Milstein B, Dietz W, Buchner D, Majestic E.  Proceedings of the 24th International System Dynamics Conference, Nijmegen, The Netherlands, 2006.

A system dynamics simulation model was developed for understanding trends in obesity in the United States.  Data on population body weight from 1971-2002 were combined with information from nutritional science and demography into a single analytic environment for conducting simulated policy experiments.  Interventions among school-aged youth and others were simulated to learn how effective new interventions would have to be to alter obesity trends; which population subsets ought to be targeted; and how long it takes for those actions to generate visible effects.  One finding is that an inflection point in the growth of overweight and obesity prevalence probably occurred during the 1990s.  Another is that new interventions to assure caloric balance among school-age children—even if very effective—would likely have only a relatively small impact on the problem of adult obesity.  More comprehensive efforts at all ages are needed to avoid the high costs and burden of disease due to adult obesity.

Hospital Surge Capacity Planning

Client:  Health Resources and Services Administration (HRSA), US Department of Health and Human Services
Manley W, Homer J, Hoard M, Roy S, Furbee P, Summers D, Blake R, Kimble M.  Proceedings of the 23rd International System Dynamics Conference, Boston, Massachusetts, 2005.
Hoard M, Homer J, Manley W, Furbee P, Haque A, Helmkamp J.  Systems Modeling in Support of Evidence-Based Disaster Planning for Rural Areas.  Intl. J of Hygiene and Environmental Health, 208: 117-125, 2005.

A system dynamics model was developed to help hospitals assess their ability to handle surges of demand during various types of disasters.  The model represents all major flows of patients through a hospital and indicates how specific responses to a surge may ameliorate bottlenecks and their potentially harmful effects on patients.  The model was calibrated to represent a specific hospital in West Virginia and was tested under three quite different surge scenarios: a bus crash, a chemical plant leak, and a SARS outbreak. Under the difficult conditions of the SARS scenario, avoidable deaths of patients awaiting emergency care could be effectively reduced by adding reserve nursing staff not in the emergency department, as might be expected, but in the overloaded inpatient wards.  The model can help hospital planners better anticipate how patient flows may be affected by disasters, and identify best practices for maximizing the hospital’s surge capacity under such conditions. 

Local Strategy for Chronic Disease Management and Prevention

Client:  a large public hospital in Washington State, under funding from the Robert Wood Johnson Foundation   
Publication:  Homer J, Hirsch G, Minniti M, Pierson M. Models for Collaboration: How System Dynamics Helped a Community Organize Cost-Effective Care for Chronic Illness. System Dynamics Review, 20(3): 199-222, 2004.

Experts agree that the U.S. healthcare system is poorly organized to care for chronic illnesses and wasteful and unresponsive to the needs of patients.  The “Pursuing Perfection” program sought to improve chronic care in Whatcom County, Washington State.  SD models focusing on diabetes and heart failure supported the planning of that program.  The models project the program’s costs and benefits over 20 years and gave its leadership the ability to do resource planning, set realistic expectations, determine critical success factors, and evaluate the differential impacts on affected parties.  Relying upon model projections, the leadership sought ways to address concerns about financial “winners” and “losers” so that all parties are willing to participate in and support the program.

Antibiotic Resistance Dynamics

Client:  Texas Department of Health
Homer J, Ritchie-Dunham J, Rabbino H, Puente L, Jorgensen J, Hendricks K. Toward a Dynamic Theory of Antibiotic Resistance. System Dynamics Review, 16(4): 287-319, 2000.
Homer J, Jorgensen J, Hendricks K.  Modeling the Emergence of Multidrug Antibiotic Resistance. Proceedings of the 19th International System Dynamics Conference, Atlanta, Georgia, 2001.

Many common bacterial pathogens have become increasingly resistant to the antibiotics used to treat them. The evidence suggests that the essential cause of the problem is the extensive and often inappropriate use of antibiotics, a practice that encourages the proliferation of resistant mutant strains of bacteria while suppressing the susceptible strains. However, it is not clear to what extent antibiotic use must be reduced to avoid or reverse an epidemic of antibiotic resistance, and how early the interventions must be made to be effective.  To investigate these questions, a small system dynamics model was developed portraying changes over a period of years to three subsets of a bacterial population – antibiotic-susceptible, intermediately resistant, and highly resistant.  The details of this model are based on a case study of Streptococcus pneumoniae, a leading cause of illness and death worldwide. 

Cocaine Use Prevalence Estimation and Policy Analysis

Client:  National Institute of Justice, US Department of Justice

Selected Publications:
Homer J.  A System Dynamics Model of National Cocaine Prevalence. System Dynamics Review, 9(1): 49-78, 1993.

Homer J. A Dynamic Model of Cocaine Prevalence in the United States.  In System Dynamics, ed. Y Barlas, Encyclopedia of Life Support Systems (EOLSS), available at Developed under auspices of UNESCO, EOLSS Publishers, Oxford, UK. 2004.

A system dynamics model reproduces a variety of national indicator data reflecting cocaine use and supply over a 15-year period and provides detailed estimates of actual underlying prevalence. Sensitivity testing clarifies the source of observed trends.  Alternative scenarios with possible policy implications were simulated and projected.  In one analysis, the model was applied to determine the potential impact of policies involving a relaxation of law enforcement.  The model suggests that a policy that eliminates both drug seizures and retail-level arrests would reduce the criminal justice load, but could lead to a large increase in cocaine use and addiction. 


Hardware Maintenance Field Service Dynamics

Client:  a major producer of diagnostic equipment used in semiconductor wafer fabrication
Publication:  Homer J.  Macro- and Micro-Modeling of Field Service Dynamics. System Dynamics Review, 15(2): 139-162, 1999.

A system dynamics model to investigate field service issues was developed for a major producer of equipment for semiconductor manufacturing. This strategic model has a broad scope and multi-year time horizon, and treats variables in an aggregate and deterministic way that is typical for such models. The high-level approach is adequate in most respets, but lacks the detail necessary to resolve a key issue regarding the impact of product cross-training on service readiness. As a result, it proved useful to supplement the strategic 'macro' model with a 'micro', OR-type model that portrays the daily queuing and assignment of service jobs. The micro model provides detailed what-if results that were used for calibrating the strategic model and may also be used for making tactical manpower decisions at the local level. Traditional OR tools may have a role to play in supporting strategic modeling efforts when important operations-level relationships are not adequately understood.  

Strategies to Improve Freight Railroad Performance

Client:  a major freight railway company (CSX Transportation)
Publication:  Homer JB, Keane TE, Lukiantseva NO, Bell DW.  Proceedings of the 1999 Winter Simulation Conference, Phoenix, Arizona.

An SD model was developed to assist the company in strategic planning.  Freight railroads in the US in the late 1990s had chronic problems with on-time service performance, which, in turn, generate costs and tie up capacity.  When capacity is already tight, train delays can lead to a vicious cycle, and in the worst case to prolonged gridlock, as occurred with Union Pacific in 1997.  Railroad cars went missing, crossings were blocked, major terminals congested, and customers factories closed, leading to customer lawsuits.  Increasing demand and shifts in demand among different lines of business (merchandise, coal, automobiles, train-to-truck intermodal) complicate the picture further.  The SD model helped the client understand how to avoid congestion problems and improve on-time performance over a three-year time horizon of increasing demand growth.  It suggested that solutions of three types were required: (1) capital solutions to add track, terminals, and equipment; (2) demand management to make seasonal adjustments and better allocate limited car capacity; and (3) operating solutions that could involve increasing the number of cars per train, establishing more reliable schedules, creating more flexibility in pick-up and delivery times, and improving productivity.     

Projecting Motorcycle Parts and Accessories Sales

Client:  a large US manufacturer of motorcycles
Publication:  Homer J.  Structure, Data, and Compelling Conclusions: Notes from the Field. System Dynamics Review, 13(4): 293-309, 1997.

For planning and strategy purposes, the company needs to be able to project parts and accessories revenues several years into the future.  These revenues are generated by shipments from the factory to dealers.  In the mid-1990s, these shipments became harder to project, as dealer incentives were phased out.  An SD model helped the client understand and anticipate better what was going on with consumer demand at the retail level.  Demand for parts is a function of several factors, one of which is the average number of miles driven by consumers, associated with wear and tear, breakdowns, and collisions.  The company was concerned, based on one imperfect type of data, that miles driven was declining rapidly and would lead to a significant contraction in parts demand.  Through a careful triangulation of other data and stock-flow logic, the model led to the conclusion that miles driven may be declining but not rapidly.  This analysis allowed the company to better project parts and accessories sales.    

Managing the Inventory of Test Items Used in Computer-Based Educational Testing

Client:  Educational Testing Service (ETS), the world’s leading developer and provider of standardized educational tests
Publication:  Homer J.  Structure, Data, and Compelling Conclusions: Notes from the Field. System Dynamics Review, 13(4): 293-309, 1997.

In the mid-1990s, ETS started a transition from paper-and-pencil testing to computer-based testing for all of its graduate-level tests.  Computer-based tests, offered on an almost daily basis at many test locations around the world and throughout the year, draw questions from a weekly pool of test items many times larger than the test itself, and that weekly pool itself gets refreshed from an even larger inventory of items.  This inventory represents a substantial investment to ETS, because the items must be written and honed for precision, reviewed for cultural and gender bias, and pre-tested on large samples of test takers.  Both item security and millions of dollars in item development costs are affected by the way in which the item inventory is managed.  An SD model was developed to project the number of available test items under different assumptions about security risk, and also to look at policies for more cost-effective inventory management.  The model revealed that in cases of higher security risk, under existing policies, the number of available items could stagnate at an unacceptably low level.  Model tests showed that a modest and safe amount of “recycling” items used in the past could neutralize this potential problem.  Moreover, the model showed that, even when security risk was not high, the policy of recycling could effectively reduce ETS item creation costs and help their bottom line.  The model analysis was presented to the ETS executive board and affected their decisions regarding test item inventory policy.     

Marketing Strategy for a New Cholesterol-Lowering Drug

Client:  a large global pharmaceutical company (Sandoz, now Novartis)
Publication:  Homer J.  Why We Iterate: Scientific Modeling in Theory and Practice. System Dynamics Review, 12(1): 1-19, 1996.

This was the first of a series of models with Sandoz dealing with the positioning of new products in various therapeutic application markets.  The cholesterol-lowering drug was a “me-too” drug entering a very competitive market.  The purpose of the model was to investigate a variety of positioning alternatives, mainly focused on pricing and marketing, to determine whether the new drug could become a winner and what it would take.  The model was capable of explaining historical prescriptions and marketing data of competitor firms, and revealed the importance of product switching, pre-marketing response, and patient compliance.

Analyzing Price Cycles in Commodity Chemicals

Client:  a large global chemical company (Dow Chemical)
Publication:  Homer J.  Why We Iterate: Scientific Modeling in Theory and Practice. System Dynamics Review, 12(1): 1-19, 1996.

An SD model was developed to explain strong repeated price cycles in the chlor-alkali chemical market.  A first version of the model was developed focusing on Europe alone.  A second, expanded, version considered the entire global market.  The model led to an improved understanding of what caused the problematic price cycle, whether it could be accurately forecasted, and whether anything could be done by Dow to dampen the cycle or to manage more effectively around it.