Complex Systems Thinking in Health

David Stephen, Craig Stephen, Luis Pedro Carmo, and John Berezowski

The concepts of systems, complexity, and chaos are appearing with increasing frequency in the health literature. Much health research and practice have, however, historically been successfully guided without evoking complexity but rather by applying a linear, reductionistic paradigm. That paradigm views reality as the sum of components that can be separated and studied as isolated entities. The reductionistic approach has been extremely successful in combating many diseases. Diseases caused by single elements, like a vitamin deficiency or a bacterial infection, can be remedied by targeting that element alone, saving countless lives, relieving much animal suffering, and improving the well-being of many. The challenges of chronic diseases, however, highlighted the limitations of the one-to-one linear model of disease causation. Lifestyle, genetic, environmental, and social factors interact in complicated ways over varying time scales that make prediction of clinical outcomes of many chronic diseases challenging and limits the value of attacking only one component cause as a management strategy. Epidemiologists began to ask if “the challenge of studying causality [can] be adequately addressed if emphasis continues to be placed on using tools and methods that are geared towards looking at... systems from a linear paradigm?” (Philippe and Mansi, 1998). Things get further complicated as our attention turns from disease to health, where health is characterized as the cumulative effect of capacities and resources derived from interacting individuals and social and environmental determinants necessary to adapt to, respond to, or control life’s challenges and changes (see Chapter 2 on defining health). The potential for additive and multiplicative interactions between determinants, synergistic and antagonistic relationships, and varying social perceptions influencing how we weigh and value these interactions make it hard to accept that health is a topic well suited to the linear, reductionist paradigm. People working in ecosystem-based management, EcoHealth, and the sustainability and health agendas are now thinking in terms of systems in which the parts and whole are inevitably connected (Waltner-Toews and Kay, 2005). To make matters even more complicated, One Health now asks us to consider interactions between different types of health for interacting species and over multiple generations that each change over time, as do the hazards and harms they experience. Health, as it is experienced in real life, is complex and messy.

A complex systems approach has been evoked as a response to calls for an alternative to the linear, reductionistic paradigm. Throughout this book, there are many instances where authors have referred to health and One Health problems as being complicated, complex, or wicked. It is hoped and hypothesized that a complex systems approach would broaden the spectrum of methods to design, implement, and evaluate One Health interventions. Yet concepts like complexity and chaos are foreign to many health scientists and “there is some looseness in how they have been translated from their origins in mathematics and physics, which is leading to confusion and error in their application” (Rickies et ah, 2007). In this chapter, we introduce complex systems and how complexity thinking can be relevant to One Health. Our goal is to lay a foundation from which interested readers can dive into the growing literature on complexity and health.

WHAT IS A COMPLEX SYSTEM?

A complex system consists of many parts interacting in a nonlinear fashion. Here, nonlinear means that a change in one part does not have a fixed effect on the whole system, rather it depends on the current state of that part and other parts in the system (Strogatz, 2015). Unlike simple, linear systems, a complex system cannot be understood by extrapolating the behaviour of the individual parts. The parts, and the organized structures they create, change dynamically, interdepen- dently, and often unpredictably over time. In this section, we will describe some phenomena that occur in complex systems, and the implications they have on how we must adapt our usual linear, reductionist approaches to encompass complex behaviour. A more extensive glossary of concepts related to complex systems, and their relation to health sciences, is given by Rickies et al. (2007).

A defining characteristic of complex systems is the notion of a feedback loop. This occurs when the interactions between parts results in a collective behaviour that feeds back into the behaviour of the individual parts, thus dampening or amplifying changes made to the system. In the former case, the system can display rigidity or adaptability in the face of a changing environment. In the latter case, small changes such as internal stochastic effects or external perturbations can be amplified to have a drastic impact on the system, such as causing a sudden transition into another regime of stability.

Another hallmark of complex systems is the existence of chaos. Since chaos is often incorrectly used synonymously with complexity, we need to clarify its meaning. Chaos refers to a well-defined mathematical notion in which two initial conditions that start arbitrarily close to each other diverge exponentially fast in time. This is colloquially known as the butterfly effect. In a chaotic system, the extreme sensitivity to initial conditions can make system change appear irregular and even random, even though the system is evolving deterministically according to some simple rules. Chaos appears in many mathematical models, even simple ones with a single variable. Chaos theory reminds us that, even in a completely deterministic system, it can be impossible to accurately predict its future state, and that apparently random behaviour need not be due to stochastic external effects.

Complex systems operate on different scales over time and space, and each scale can exhibit entirely new properties that cannot be extrapolated from the behaviour on smaller scales (Anderson, 1972). This is due to the principle of emergence, which arises from the intricate network of interactions between the many parts of a system. Disease systems, for example, can be simplified as a multilevel nested hierarchy (Ceddia et al„ 2013). The “ground-floor” level is where the disease occurs. Below that is the level of individual decision and relationships affecting the day-to-day small-scale actions and relationships that influence disease dynamics. Above is the level of collections of individuals and institutions that enable or dissuade actions and decisions. Above that is the biophysical level where ecosystem processes influence how the other constituent parts of the system interact. The level at which one encounters (or studies) such a disease system will affect the perspective one has of the system. This means that when viewed from different perspectives or disciplines at different times and scales, the “same” complex system can be described differently.

A comprehensive understanding of a health outcome requires us to look at multiple variables interacting across all levels and across different spatio-temporal scales. This seems an overwhelming task, one to which people are increasingly evoking complexity theory as help. There is. however, no single “complexity theory” per se. The term complexity theory refers to several fields of study which all aim to (i) understand which obstacles prevent us from predicting the evolution of a system, such as those described above, and (ii) find ways around these obstacles (e.g. using mathematical or statistical methods). A complexity-based approach involves questions different than asking: “Does pathogen A cause disease B?” or “What risk factors are associated with the transmission of infection?” (Pearce and Merletti, 2006). Instead, it is better used to ask, “Are there circumstances where certain subpopulations are more vulnerable to a disease?” or “Are there situations where surveillance resources would be more likely to detect an emerging issue?” or “Which upstream intervention should we target knowing that there are many intervening variables between the intervention and the health outcome that could modify its impact?”

The interdependence of humans, animals, and their shared environments, the co-evolution of their interactions, the emergent properties of their behaviours, feedback loops within the system which enable or constrain further behaviours, the networked nature of relationships, and the different scales of socio-ecological systems, all point out that One Health operates in the sphere of complex systems. There are implications to thinking about complexity in One Health (based on Preise et al., 2018). Firstly, it shifts our attention from components of the system to the system as a whole. In a systems approach to a problem, the emphasis shifts from the parts that make up the system to their interrelationships (Pohl and Hadron, 2008). This requires us to pay attention to organizational processes, connections, and emergent behavioural patterns of the system. This shift asks us to spend less time looking for causal pathways and more time characterizing relationships and interactions that influence patterns of system behaviour to facilitate our understanding of how systems transform and how emerging characteristics arise. Secondly, adopting this perspective requires the use of different tools and methods adept at capturing and characterizing relationships such as network analysis, participatory systems analysis, and transdisciplinary methods. These methods need to be adaptable, they should cultivate social learning, and they should favour synthesis over isolated analysis. The tools and methods need to be able to capture spatial and temporal dynamics, be attentive to surprises, and identify critical thresholds and tipping points. System dynamics models (Keeling and Eames, 2005), agent-based modelling, and time-series analysis are candidate methods. Finally, we need to be aware that the application of our tools and methods could affect the system and that some boundaries for our investigations will need to be imposed or constructed, which in turn will influence what we measure.

 
Source
< Prev   CONTENTS   Source   Next >