Session 1: Introduction to the CAS approach to Health Systems


Session 1: Introduction to the CAS approach to Health Systems


Professor David Peters (Johns Hopkins University and Future Health Systems Consortium)

Health systems in developing countries are usually comprised of highly heterogeneous groups of actors (e.g. many types of health care providers, managers, policy-makers, patients, regulators, etc.) intervening at multiple levels through a variety of services and functions. The interconnectedness of actors and their dynamic interactions across the health system closely resemble the characteristics of Complex Adaptive Systems (CAS) (Tan et al 2005; Rickles et al 2007; World Bank 2007; de Savigny & Adam 2010).  Health systems are described as CAS because in addition to being comprised of many interacting components, they have the capability to self-organize, adapt, or learn from experience. The interactions of system components are typically complex and change in non-linear ways over time, and are not easily controlled or predictable in detail.  They also result in unintended effects or paradoxical behavior. It is not unusual for CAS to show little response to many attempts to control it, or to change suddenly when a tipping point is reached (Gladwell 2000).  For example, many high cost health investment projects have had little impact on people’s behavior or health status, in contrast to sudden changes that can occur in public opinion about smoking bans or in the demand for contraception. 


People’s understandings of systems that are actually CAS are often over-simplified or erroneous.  This leads to poor policy and management decisions for decision-makers who cannot control such systems through conventional means, often while being vulnerable to sudden changes in public opinion (Dorner 1997).  Many CAS phenomena are relevant to the pathways of health systems, including path dependency, feedback loops, scale-free networks, emergent behavior, and phase transitions or tipping points (Paina & Peters 2010). Unanticipated consequences related to interventions in health systems can also be extremely, including: 

  • populations that become marginalized as a result of new or reorganised health services (e.g. many health service interventions fail to reach those most in need); 

  • populations that become marginalized as a result of new or reorganised health services (e.g. many health service interventions fail to reach those most in need); 

  • failure to allow for disorganized markets, such as occur for example where institutions are not in place to assure the quality of drugs or health services, where informal payments undermine formal financing mechanisms, or where well-intentioned community health worker programs result in the creation of a new group of informal sector providers without effective regulation; 

  • narrowing of the range of services, which can result when projects or priority programs displace other essential services, or where a failure to address incentive structures leads to a decline in financially less attractive services such as those for preventive care; 

  • unsustainable strategies, for example when islands of excellence are created that are dependent on the availability of short-term external resources; 

  • crowding out of key actors, which occurs when a project displaces existing service providers rather than integrating or complementing their work and;

  • failure to recognise potential risks to public goods, for example the emergence of resistant pathogens where initiatives result in the widespread and poorly regulated use of specific drugs or practices.

The lens of CAS opens up a deeper understanding of how to effect change in health systems, including the pathways for increasing and sustaining coverage of effective interventions.  It proposes a shift in thinking from the current models around scaling up health services, which revolve around linear, predictable processes, to models that embrace uncertainty, non-linear processes, the uniqueness of local context, and emergent characteristics.  

Understanding the pathways for change in a CAS has much to offer.  The blueprint approaches commonly found in global health initiatives, with an emphasis on detailed initial planning and inflexible designs, are not a good match for addressing the adaptive properties of dynamic pathways for expanding health services.  CAS models may also identify opportunities to create more effective health services or reach marginalized target groups, such as by identifying critical points for phase transitions, emergent behaviors, or using growing or untapped networks to spread effective practices, in addition to anticipating unintended consequences.  Frequent analysis of data on the implementation of health services and their effects is a common approach for effectively understanding and intervening in a health system considered as a CAS. 

There are many areas of research that might benefit from using a CAS approach: 

  1. Can we characterize health system dynamics in ways that will help change health systems to better performance, and learn how to avoid low performance equilibria or system instability (e.g. disorganized health systems)? 

  2. Can we construct “quasi-experiments” or take advantage of ongoing change in health systems to examine how to change health systems, including identification of stable, unstable, and meta-stable states with different levels of performance? 

  3. Can we identify the key dynamic properties of health systems to understand how to better influence them, given the different pathways followed across different settings?  For example, can we better understand the nature of relationships and identify key points of change through the tracking of: 

  • Health Systems Performance (multiple criteria based on health status, equity and efficiency of health status);

  • System complexity (e.g. heterogeneity of actors, institutions, degree of fluctuations);

  • Degree of decentralization of decisions (over health behaviors, service quality and use); 

  • Types of perturbations to health systems; 

  • Types of control parameters (e.g. Financing, Public information, Provider influences, Organization of delivery); 

  • Degree of adaptability in the health system (esp. information flows and feedback, and responses to perturbations and controls) 

  1. Can we identify ways to encourage: 

  • Productive emergent properties (such as those related to self-regulatory measures over safety, quality, and/or pricing of health services) 

  • Diffusion of information to promote health behaviors &/or health goods and services

  • Firewalls to protect against negative chain reactions (or system cascades) that would lead to loss of trust in health markets, dissemination of harmful health practices, or epidemic disease 

  • Systems to ensure information is useful for accountability or to provide early warning information 

To address these issues scientists will need to work more closely across disciplines and with policy-makers to provide better knowledge for decision-making in policy and program planning, implementation, monitoring, and evaluation.  Although complex systems raise particular management and research challenges, there are already a wide variety of theories and methodologies that have been developed to address them, including (Helbing 2009): 

  • Large-scale data mining 

  • Network analysis 

  • Systems dynamics 

  • Scenario modelling 

  • Sensitivity analysis 

  • Non-equilibrium statistics (physics) 

  • Non-linear dynamics and chaos theory 

  • Systems theory and cybernetics 

  • Catastrophe theory 

  • Statistics of extreme events 

  • Theory of critical phenomena 

  • Agent-based modelling 

Description of CAS concepts  

Listen to an introductory podcast


References

deSavigny D, Adam T. Systems thinking for health systems strengthening. Geneva, Switzerland: Alliance for Health Policy and Systems Research, World Health Organization 2009.

Craig,Peter, Paul Dieppe,Sally Macintyre, Susan Mitchie,Irwin Nazareth, Mark Petticrew. (2008). Developing and evaluating complex interventions: the new Medical Research Council guidance. BMJ 337: 979-083 (pdf)

Dorner D. The Logic of Failure: Recognizing and Avoiding Error in Complex Situations. New York: Basic; 1997

El-Sayed, Abdulrahman, M., Peter Scarborough, Lars Seemann4 and Sandro Galea. 2012. Social network analysis and agent-based modeling in social epidemiology. Epidemiologic Perspectives & Innovations 9(1).

Gladwell M. The Tipping Point: How Little Things Can Make a Big Difference. London: Abacus; 2000.

Helbing D. Systemic Risks in Society and Economics: Santa Fe Institute; 2009.

Medical Research Council (2008). Developing and evaluating complex interventions: new guidance. (pdf)

Paina L, Peters DH. Pathways to Scaling Up Health Services in Developing Countries. Health Policy and Planning.. 2011.

Rickles D, Hawe P, Shiell A. A simple guide to chaos and complexity. Journal of epidemiology and community health. 2007;61(11):933-7.

Subramanian S, Naimoli J, Peters DH. Scaling Up and the Millenium Development Goals. Manuscript under review. 2010

Tan J, Wen JH, Awad N. Health Care and Service Delivery Systems as Complex Adaptive Systems. Communications of the ACM. 2005;48(5).

Uvin P. Fighting hunger at the grassroots: Paths to scaling up. World Development. 1995;23(6):927-39

Wolcott, Sara J (2012). Towards Research Strategies in Future Health Systems: Exploring other sectors approaches to complex dynamics. Institute of Development Studies 

World Bank. Healthy Development: The World Bank Strategy for Health, Nutrition, and Population Results. Washington, D.C.: World Bank; 2007.

Xiao, Yue, Kun Zhao, David M Bishai & David H Peters (2013). Essential drugs policy in three rural counties in China: What does a complexity lens add? Social Science & Medicine 93:220-228 (abstract)


Back to Main Page

Last modified: Monday, 27 July 2015, 6:29 AM