• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of amjphAmerican Journal of Public Health Web SiteAmerican Public Health Association Web SiteSubmissionsSubscriptionsAbout Us
Am J Public Health. 2010 July; 100(7): 1161–1163.
PMCID: PMC2882409

Systems Science: A Revolution in Public Health Policy Research

The new systems science approaches emerging in public health research are not new at all; they have a track record earned over several decades in other disciplines, such as physics, operations research, economics, engineering, and, more recently, systems biology. At their core, systems science methodologies are designed to generate models, or simplified versions, of reality. By replicating the real world in important ways—simplifying where possible while retaining the critical aspects relevant to the problem under study—we can better understand the structural complexity of real-world problems that results from the interaction of specific phenomena and their environments. Systems science approaches have been used to address wide-ranging topics such as wildfire control, overfishing, decline of ancient civilizations, climate change, and terrorism networks. A major reason for their recent adoption in the public health arena is a growing recognition of their utility for addressing the complex problems rampant in public health generally1 and in specific domains (e.g., neighborhood effects on health and obesity).2,3 A small but growing number of studies have employed systems science methodologies to understand and address public health problems such as pandemic flu,4 HIV/AIDS,5 diabetes prevalence,6 and heroin markets.7 A few researchers have been using systems science to further tobacco control for nearly a decade.8,9


By repurposing systems science methodologies for public heath research, we can add tremendously to the depth of our existing methodological and problem-solving toolbox to help us break through some large and intractable problems, such as the smoking and obesity epidemics. Traditional methodologies based in reductionism have been, and will continue to be, extremely useful for gaining new knowledge about direct causal relationships and for understanding components of problems. Mathematical modeling and other systems science approaches can add to this arsenal of analytic tools. These approaches are specifically geared to identifying, understanding, and quantifying nonlinear relationships among system components, time-delayed effects, bidirectional relationships between component parts (i.e., feedback loops), and emergent phenomena (i.e., properties of the system that emerge from the individual constituents of the system without being attributable to any given component). Systems science approaches generate, in a systematic and informed way, plausible scenarios of what the future might look like under various specified conditions.

The methodological divide that is often posed between a reductionist paradigm and a systems paradigm is a false dichotomy; we advocate the use of systems science approaches in concert with, not in lieu of, other methods. Systems science approaches refer to a wide range of methodologies, and there is no single best method. Rather, each methodology is useful for exploring certain types of questions. In cases in which more than one approach is appropriate, much can be learned by comparing model structure, assumptions, boundaries, and results across different methodologies. Observing where the different approaches converge can help establish confidence in those outcomes, whereas observing where the models diverge can alert the user to important differences in model structure, parameterization, data used to inform the model, and other model attributes that should be explored to gain additional insight.


Here are some of the most powerful uses for systems science methodologies along with some illustrations of how they can be used for tobacco control research.

Systematically Synthesize the Data We Already Have

Systems science has the potential to make sense of volumes of data from many disparate disciplines. It allows us to synthesize meaning from a broad evidence base by simultaneously exploring relationships among different subareas of the field being considered. For example, modeling might be the most advantageous way to get an idea of how quitlines fit within the broader arsenal of tobacco control polices by synthesizing information from a variety of areas.10 This kind of synthesis can be thought of as getting an overarching view of the existing knowledge about the problem.

Expose Gaps in the Existing Body of Knowledge

Modeling is very useful for determining what information is missing and how critical it is for understanding the bigger picture. For example, if you wanted to model the prevalence of tobacco-attributable deaths in a county as a function of tobacco policies—for example, taxes, clean indoor air laws, or media campaigns—and you did not have good data on how much impact media campaigns had on smoking prevalence in that county, the model outputs could be tested to determine how sensitive they are to the specified inputs (effects of media campaigns). The larger the range of uncertainty about the outputs, the more important it is to seek the missing input. Lacking county-specific data, you could do the next best thing: estimate the media's effect on the basis of studies in other counties. Adjusting the data to reflect the demographics of the county being examined could further improve these estimates. The uncertainty of using nonspecific data would be reflected in the output.

Inform Research Priorities

Identifying data gaps and determining how critical they are to the research questions being posed can be very useful for informing research priorities. By assessing which model inputs create the most (and least) uncertainty in model outputs, we can get an idea of which research investments are likely to have the biggest payoff for understanding the big picture. In general, inputs contributing the most to model uncertainty, if better specified, have the greatest potential to improve the accuracy of model outputs. However, improving model accuracy may not be the only or primary concern, and factors not captured in the model may be important. Therefore, human judgment should never be abdicated in deference to the model.

Identify Leverage Points in the System

Which perturbations in the system (i.e., policy changes) lead to the most desirable outcomes (i.e., the highest probability of the most desired results) and over what time frame? What unintended consequences of a policy or intervention could occur? Often decisions about how to intervene are made on the basis of data evaluation that excludes the larger context. For example, a policy designed to reduce smoking among youths via media messages may not take into account the reaction to the policy on the part of the tobacco industry. By quantifying complex relationships in a mathematical model, plausible scenarios that could result from policy interventions can be simulated. Virtual experimentation involves generating a variety of simulations on the basis of differing assumptions and model inputs (e.g., implementing a policy or intervention under various conditions). For example, the impact of raising cigarette taxes on youth uptake could be modeled under conditions in which the tobacco industry counteracts the tax with its own strategies, such as a new marketing campaign for smokeless tobacco products.

Outcomes of such campaigns could be modeled under varying assumptions, such as differing levels of effectiveness in convincing youths to purchase a new product or in ultimately causing them to switch over to cigarettes or become dual users of both cigarettes and smokeless tobacco. Keeping the range of uncertainty in mind, the virtual experimenter can evaluate the tradeoffs associated with various decisions; the model can even help determine the best combination of strategies to employ over what time frame for a given set of model assumptions.


For tobacco control and public health researchers to get the most benefits out of systems science approaches, there must be some coordination of effort. The National Institutes of Health (NIH) has already begun to encourage the use of systems science methodologies in several funding opportunity announcements. The more the methods are understood and the more investigators practicing the methods, the more NIH and other funding agencies will accept and ultimately fund them, and the more journal editors and reviewers will appreciate their value and accept them for publication. The appetite for systems science approaches will undoubtedly be piqued in many readers of this issue of the Journal.

What can and should readers do to learn more about systems science and its potential for addressing public health problems? NIH and the Centers for Disease Control and Prevention have led a variety of activities designed to foster the revolution of systems science in public health. In 2007 the Office of Behavioral and Social Sciences Research and the Centers for Disease Control and Prevention produced a series of videocasts that serve as a primer on systems science and health.11 Readers can also continue to learn about the potential for systems science through theme issues like this one and the 2006 theme issue of the Journal on systems thinking12 as well as National Cancer Institute Monograph 18.13 Applications can be submitted to NIH using systems science approaches, and researchers can serve on NIH review committees and journal review boards to help such entities be more accepting of these methods. Articles of interest, training opportunities, funding opportunity announcements, and a variety of other information are posted to the Behavioral and Social Sciences Systems Science Listserv on an ongoing basis. (Please contact the list owner, P. L. M., to subscribe to this service.)


The authors gratefully acknowledge those who helped to start the revolution of systems science in public health, including Bobby Milstein (Centers for Disease Control and Prevention), David Abrams (American Legacy Foundation), Scott Leischow (University of Arizona), and David Levy (Pacific Institute for Research and Evaluation).

Note. The findings and conclusions in this editorial are those of the authors and do not necessarily represent any official position of the Office of Behavioral and Social Sciences Research, the National Cancer Institute, or the National Institutes of Health.


1. Mabry PL, Olster DH, Morgan GD, et al. Interdisciplinarity and systems science to improve population health: a view from the NIH Office of Behavioral and Social Sciences Research. Am J Prev Med 2008;35(2):S211–S224 [PMC free article] [PubMed]
2. Auchincloss AH, Diez Roux AV. A new tool for epidemiology: the usefulness of dynamic-agent models in understanding place effects on health. Am J Epidemiol 2008;168(1):1–8 [PubMed]
3. Hammond RA. Complex systems modeling for obesity research. Prev Chronic Dis 2009;6(3):A97. [PMC free article] [PubMed]
4. Longini IM, Jr., Nizam A, Xu S, et al. Containing pandemic influenza at the source. Science 2005;309(5737):1083–1087 [PubMed]
5. Morris M, Kretzschmar M. Concurrent partnerships and transmission dynamics in networks. Soc Networks 1995;17(3–4):299–318
6. Jones AP, Homer JB, Murphy DL, et al. Understanding diabetes population dynamics through simulation modeling and experimentation. Am J Public Health 2006;96(3):488–494 [PMC free article] [PubMed]
7. Hoffer LD, Bobashev G, Morris RJ. Researching a local heroin market as a complex adaptive system. Am J Community Psychol 2009;44(3–4):273–286 [PubMed]
8. Levy DT, Chaloupka F, Gitchell J, et al. The use of simulation models for the surveillance, justification and understanding of tobacco control policies. Health Care Manag Sci 2002;5(2):113–120 [PubMed]
9. Tengs TO, Osgood ND, Lin TH. Public health impact of changes in smoking behavior: results from the tobacco policy model. Med Care 2001;39(10):1131–1141 [PubMed]
10. Levy DT, Graham AL, Mabry PL, et al. Modeling the impact of smoking-cessation treatment policies on quit rates. Am J Prev Med 2010;38(3 Suppl):S364–S372 [PMC free article] [PubMed]
11. National Institutes of Health, Office of the Director, Office of Behavioral and Social Sciences Research Videocasts. Available at: http://obssr.od.nih.gov/training_and_education/videocast/videocast.aspx#ssh. Accessed April 4, 2010
12. Northridge ME, McElroy K, editors. , eds Systems Thinking. Am J Public Health 2006;96(3, theme issue):398–576
13. National Cancer Institute Greater Than the Sum: Systems Thinking in Tobacco Control Bethesda, MD: US Department of Health and Human Services, National Institutes of Health, National Cancer Institute; 2007. NIH publication 06-6085. Tobacco Control Monograph Series No. 18. Available at: http://cancercontrol.cancer.gov/tcrb/monographs/18/index.html. Accessed April 4, 2010

Articles from American Journal of Public Health are provided here courtesy of American Public Health Association
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Cited in Books
    Cited in Books
    PubMed Central articles cited in books
  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...