Skip to main content

Table 1 Aspects of complexity to consider in developing guidelines on public health and health system interventions (adapted from Petticrew and colleagues [5] and Rehfuess and colleagues [6])

From: When complexity matters: a step-by-step guide to incorporating a complexity perspective in guideline development for public health and health system interventions

Aspects of complexity Description Example
Interactions between intervention components Interventions may include multiple interacting components which may have synergistic or dysynergistic effects on the system. Parenting interventions to prevent child maltreatment may include multiple components relating to parenting knowledge, skills and parental mental health to bring about changes in the family environment and parent-child relationships [7].
Interaction of interventions with context Interventions – and their components – may be context-dependent, i.e. their effectiveness, feasibility and acceptability may be affected by the epidemiological, socio-economic, socio-cultural, political, legal and other characteristics of a given context. Giving corticosteroids to women at risk of pre-term delivery can effectively reduce the risk of fetal and neonatal deaths in hospitals with special care in high-income countries; in countries without such special care hospitals antenatal corticosteroid therapy may do more harm than good [8].
Dynamism Systems evolve and change over time as a result of interactions among diverse agents. School teachers, staff and students constantly change within schools [9].
Adaptivity and co-evolution Interventions may influence the context of implementation (directly or indirectly). The entire system adapts and responds in expected or unexpected ways. The interventions themselves also change in response to system changes. Regulations to ban smoking in public places or to prohibit the sale of tobacco products with certain characteristics may affect individual consumption; manufacturers may reformulate tobacco products as a response. This may further change how regulations are formulated or implemented [10].
Emergent properties Intervention effects may emerge from self-organisation among the interacting agents. Herd immunity is an emerging effect of human papillomavirus vaccination of a sufficient percentage of the population [11].
Non-linearity and phase changes Interventions may demonstrate effects once they have reached a certain scale. Community sanitation interventions first need to reach thresholds in the order of 60% or higher, to optimise health and nutrition gains [12].
Feedback loops Interventions comprised of different components can produce feedback loops reducing the overall effect (negative), or conversely, enhancing the effect beyond what might be expected (positive). Interventions to increase the availability of healthy foods promote healthy diets, which further enhance the need for healthy foods (positive feedback loop) [13].
Multiple (health and non-health) outcomes and dependencies Interventions such as those involving multiple components often impact a large number of health and non-health outcomes and involve complex causal pathways. In addition to the direct effects of alcohol advertising restrictions on consumption, such restrictions may also affect non-health outcomes, such as spending on alcohol, risk behaviours, and social norms around consumption [14, 15].