One step back toward the future of health promotion: complexity-informed health promotion

The scholarship and practice of health promotion is ever-changing—fortunately, and our Journal attests to it. Early in 2019, I had the honor of presenting at a joint seminar held between the UNESCO Chair Global Health & Education and the European Forum for Health Promotion Research in Denmark, Copenhagen (Keshavarz Mohammadi, 2019). I outlined the key steps for shaping the future of Health Promotion, and suggested taking one step back, redefine failures through the lens of Complexity Science and then move forward—with the solutions generated from this new understanding. I will elaborate on my insights relating to complex adaptive systems (CAS) theory in this editorial. More than three decades since health promotion was codified in the Ottawa Charter, the health promotion field still faces the chronic problem of less-than-successful interventions (Green et al., 1986; WHO, 2009; Keshavarz Mohammadi et al., 2010a,b). Health promotion is also challenged by the replicability of interventions that are considered Best Practices. We still face difficulties explaining unforeseen impacts and side effects (Gugglberger, 2018), delayed effects, no effects or effects not proportionally relevant to the size or sophistication of the intervention. Explanations for this include the complexity of change, and resource limitations, but still many issues in explaining the less than optimal effects of health promotion remain unresolved. Albert Einstein said that ‘Figuring out how to think about the problem’ had been most helpful for development of his theory of relativity (Sanders, 2003). Perhaps it is time for the health promotion community to find alternative ways of thinking about its problems. I would like to propose ‘complexity science’ as an alternative new way of thinking about our problems and their solutions. Several authors [e.g. (Hawe et al., 2004; Keshavarz Mohammadi et al., 2010a,b)] concur so as to add to the understanding of health promotion development and implementation over the last decade. At the individual and population levels health and disease are generally determined by an array of individual (biological and cognitive) and environmental (including social) determinants or factors. Some of these factors are ‘by-products’, or an ‘epi-phenomenon’, resulting from other determinants (Kawachi and Subramanian, 2014). Conventionally, these links between causes and their impact on health are considered simple, independent and linear. This understanding has led to deterministic, predominantly one-to-one relationships between ‘causes’ and ‘effects’ (Eckersley, 2015). Using Karl Popper’s metaphor of ‘Clocks and Clouds’, health is seen as a clock with only a few moving and changing independent parts in an ordered and predictable circle—rather than being seen as a dynamic changing cloud that co-changes as other components change. This simplistic and conventional clock approach has some key pitfalls and shortcomings. From the complexity science perspective, as well as from the more empirical evidence base, the problem with this approach is that it ignores the networked and multi-layered nature of determinants and their emergent multipath influence on health and wellbeing (Helbing, 2013; Eckersley, 2015; Hawe, 2015). The network structure of links in complex systems implies that one health determinant can simultaneously influence more than one other factor directly or indirectly, so there are multiple potential paths of cause and effect, not a single path. Another important point is that interactions or influences between factors are often mutual and reciprocal which makes them change together, or ‘co-vary’. Sometimes one factor can mediate, weaken or strengthen the effect of other factors. As a result, change in one part of the system can cause often non-linear and unpredictable changes in other parts. From this perspective, it is also the sequence and combination of determinants of health or conditions of the target individual or settings which determines health status, not only their mere existence. A common example is that not all smokers develop disease (such as lung cancer), their emergence depending on risk factors and productive factors and their sequences and strength and the way they suppress/boost the effect on health. The simplistic understanding of health has been challenged by other researchers. For example, Eckersley (Eckersley, 2015) criticizes the narrow focus on the role of income inequality as one of the key Social Determinants of Health in youth wellbeing. He refers to research that shows no socio-economic gradients in youth’s psychological wellbeing, but higher levels of mental health problems among the well-off in some developed countries. He suggests a stronger effect of ‘youth culture’ as another determinant of health. A multi-level data analysis of European countries by Zagorski et al. (Zagorski et al., 2014) showed that although there was a general simple correlation between income inequality and some health indicators, after controlling per capita GDP, national inequality did not reduce health or wellbeing. In other words, in rich countries the causal link between income inequality and lower health status was suppressed. Consequently, they claimed that suggested inequality reduction interventions are unlikely to benefit health in rich countries. One might conclude that the complex nature of health, and the networked structure of its component determinants of health often escape analysis (Helbing, 2013; Eckersley, 2015; Gallie, 2019). As solutions follow problem identification, current solutions do not capture the dynamic interdependencies of determinants of health which leads to less than successful health promoting interventions. The evidence shows the need to deeply understand and approach health as a complex phenomenon. From complexity science perspectives, health can be defined as an emerging complex product of systemic interplay of many continuously co-changing ‘bio-cognitive-social-techno-environmental factors’. These are influenced by initial status, life course changes and experiences, as well as unplanned events. As a result, health status would emerge as the collective effect—not easily predictable. This suggests we need to not only identify determinants of health or disease, but also the processes and the exact causal relevant path—ensuring they are tailored to each individual setting. Fortunately, new multidisciplinary research in health is adopting this insight, particularly in obesity research, approaching it as a complex health problem (Butland et al., 2007; Stok et al., 2017). Advocating shared responsibility and accountability for health which calls for multisectoral action toward health also exhibits better acknowledgement of the complexity of health (Keshavarz et al., 2019). Individuals live in families, groups and communities—environments in which they create meaning. These settings are social CAS (Cohen and Axelrod, 2000; Plsek and Greenhalgh, 2001; Keshavarz Mohammadi et al., 2010a,b; Barasa et al., 2017). However, reviewing health promoting settings interventions [and what is becoming known as the complex intervention movement (Clark, 2013)] reveal an underlying simple systems approach and mindset toward settings. For example, the complex intervention terminology suggests that more complex problems need more interventions at more levels. Yet, it ignores that nonlinear characteristics of CAS means that even a single simple intervention can produce major changes (i.e. effect) (Rogers, 2008), while a major cause, might produce little change. This non-linearity has been experienced in some multicomponent, expensive and sophisticated health interventions (Hawe, 2015). Additionally, attempts related to ‘standardizing the form’ in interventions, overlook the self-organizing characteristics of a CAS (Oakley et al., 2006). Evidence is mounting that embracing self-organization is not only possible, but also can lead to improved results (Joyce et al., 2017). Expecting a system level change by changing one component of the system, or considering change in one component, equal to whole system level change is a misconception. We see this in many healthy settings approaches such as Health Promoting Schools (Yazdi-Feyzabadi et al., 2018), Health Promoting Hospitals (Johnson and Baum, 2001), and Health Promoting Workplaces (Motalebi et al., 2017). In other words, the assumption was that implementing some health promotion programs in settings would make them health promoting systems. Foster-Fishman and Behrens (Foster-Fishman and Behrens, 2007) argue that in these systems we strive for knowledge and skills to identify leverage points—by changing one part, the whole system would change and also the mechanisms that sustains the change. We should accept that there is no ‘one size fits all’ solution, so agreeing that ‘what works for whom in what ways’ may be correct as advocated by realist evaluation (Pawson and Tilley, 1997). For health promotion, let’s consider the characteristics of CAS, in terms of their structure and function. Complexity science defines CAS as a system made of diverse interactive, interdependent, autonomous learning components (agents), located in a networked structured and rule-based system with blurred boundaries (Keshavarz Mohammadi et al., 2010a). In these systems, the collective behavior emerges as the outcome of interaction of all agents and their autonomous self-organized adaptation to their neighboring agents. So a change in some agents can alter the whole system outcome. For example, as has been reported by Bahraminejad et al. (Bahraminejad et al., 2015), changing the members of joint committees, can influence the outcome of intersectoral collaboration dramatically. This adaptive dynamic often makes the whole system behavior and the outcome of any intervention uncertain and unpredictable, as it ‘depends’ on different conditions. Hence, non-complexity informed approaches set unrealistic expectations of predictability of the intervention-outcome link, and leads to adopting a deterministic approach and more likely sets up many health promotion interventions to fail. Barriers to developing sustainable and effective system level change can be identified in a uniquely different way through the utilization of insights from complexity science, particularly CAS theory (Keshavarz Mohammadi et al., 2010a,b). Any social change—including health promotion intervention—whether it is a new idea, lifestyle, service, program or policy would fail if it is perceived as ‘inadaptable’ by target individuals and settings. More adaptability means more adoption of the intervention and more sustainability. CAS, as the name implies, can morph and realign to new contexts and realities. But, for successful adaptation, there should exist conditions including autonomous action, distributed flow of information, a set of simple rules, timely feedback loops, and an interaction network to allow effective communication. A pioneer application of complexity science, found that barriers for successful implementation of Health Promoting Schools matched very well with variables and parameters that would be established through the lens of complexity science (Keshavarz Mohammadi et al., 2010b). This study re-defined barriers from ‘lack of adequate interest among school settings’ to Limited freedom of self-organization of schools to the concept of health promoting schools. Suboptimal flow of information within schools, between schools and the health sector about the concepts and existing experiences of other schools. Not entirely sufficient interaction within health promoting schools and between schools and health sector. Delayed or insufficient re-enforcing feedback loops about schools' health performance. Insufficient rules and incentives to support schools to become health promoting schools. Limited time for producing effect or poor timing for introducing change. A recognition of the complex nature of (the determinants of) social change and health promotion requires a paradigm shift. The adoption of complexity science perspectives would establish such a shift. Green (Green, 2006) and Resnicow and Page (Resnicow and Page, 2008) have called for such a paradigmatic change. I would like to propose adopting a ‘complexity-informed health promotion’ approach as a ‘fourth public health revolution’. The three previous revolutions included the biomedical approach, psychological approach and social approach (Keshavarz Mohammadi, 2013; Beaglehole and Bonita, 2004). The challenge is developing complexity-informed theories, logic models, tools and skills so as to empower us to manage the complexity of our task. Scholarly reflection on complexity science and CAS has been offered in the past by the American Journal of Public Health (Green, 2006), Social Science and Medicine (Tenbensel, 2013), and the American Journal of Community Psychology (Foster-Fishman and Behrens, 2007). In spite of a dozen or so complexity papers (McQueen, 2000; Kremser, 2011; Alvaro et al., 2011; Van Beurden et al., 2013; Tremblay and Richard, 2014; Fisher et al., 2014; Bilodeau and Potvin, 2016; Turunen et al., 2017; Roussy et al., 2019) featured in Health Promotion International (HPI) it seems that the critical mass of scholarship and development in the health promotion domain is trailing behind. HPI calls for the submission of significant pieces of work in methodology, theoretical development, and application of CAS and complexity science to contemporary health challenges. Obesity research seems to have embraced these perspectives, while other health promoting settings and fields would benefit from a complex health promotion lens. Our journal is ready to publish more complex and more complexity papers. These may well lead to curated streams of complexity science at the next IUHPE global conference (in Montreal, 2022) and the tenth WHO Global Conference on Health Promotion. The time has never been better than now. I would like to thank HPI for providing me with the opportunity to reflect on my journey in complexity science which began in 2003 in Australia. I thank Prof. Penny Hawe and Prof. Evelyne de Leeuw for helpful comments on earlier versions of this manuscript.