��������������������������
along the same line of thought , while perhaps suggesting additional nuance . Considerable debate could be focused on the meaning of simplicity , but it seems useful to turn 180 degrees and ask how complex do we need things to be ?
Traditionally the answer to such a query in medicine and public health has been to simplify as much as possible , to get as close as possible to the tight control seen in laboratory sciences . Thus , the randomized clinical trial or community experiment , trying to make things as simple as possible through control and exclusion , are seen in public health and medical science as the gold standard . However , there is increasing reason to believe that some of the benefits of the simplicity that comes from the tight control of extraneous variables may be illusory , and there is increasing discussion of the limitations of the knowledge that come from such methods ( Ioannidis , 2005 ).
This is not just a theoretical problem , but casts shadows on many areas of both public health science and practice . We suggest that this is particularly true for research that is focused on the “ social determinants of health ( SDOH )” ( Galea , 2007 ; Kaplan , 1985 ; Kaplan , 2004 ; Kaplan , Everson , & Lynch , 2000 ; WHO Health Commission on Social Determinants of Health , 2008 ). With their wide footprint on health on other determinants of health such as behavior , SDOH present paradigmatic case studies of tightly interrelated and causally tangled problems , far beyond the historic concepts of the web of causation , and not easily subject to a dissecting out of independent causes , or experimentation . They are , in short , complex problems not easily suited to either the rarified atmosphere of randomized clinical trials or community experiments — their complexity simply doesn ’ t easily admit to such approaches .
The increasing realization of the importance of SDOH and the growing realization that a full understanding of SDOH requires an understanding of the role of social determinants as complex systems , creates a conundrum : should we ignore the SDOH because of their complexity , or should we challenge the conventional wisdom that values reductionism and the analytic armamentarium , much of which attempt to mimic experimental designs , that accompanies it ?
Fortunately , starting in the 1940s a set of tools , representing the influences of systems theory , cybernetics , and artificial intelligence began to be developed to deal with complex systems , and they are increasingly being applied to the study of population health and health disparities ( Yoav & Kuh , 2002 ). Generically referred to both as “ systems science ” and “ complex systems ” approaches , the toolkit for using such approaches is now available to public health researchers and such tools are beginning to be used to study the impact of SDOH on health and health disparities .
In what follows , we will briefly describe some of the evidence that SDOH are related to health outcomes , and then describe the complex , multi-level , and multiscale pathways that characterize two social determinants of health ( socioeconomic position , neighborhood characteristics ). We will characterize some of the challenges that this complexity presents to our analysis and understanding , describe an approach to understanding complex systems using agentbased models ( ABMs ), and then illustrate how ABM can be used , ��������� , to address counterfactuals about the role of education and neighborhood characteristics in racial / ethnic disparities in body mass index ( BMI ). We will close with some thoughts on the strengths and limitations of such an approach , and promising areas for their use .
89