When are complex interventions 'complex'? When are simple interventions 'simple'?

Abstract

The issue of complexity, and how one evaluates complex interventions, remains a key one for health service and public health researchers. The underlying question—what makes complex interventions complex?—is particularly interesting, and it has implications for the sorts of research we do. The MRC guidance1 provides one detailed answer: complexity resides (among other things) in the number of interacting components; the number and difficulty of behaviours required by those delivering or receiving the intervention; the number of groups or organizational levels targeted by the intervention; the number and variability of outcomes; and the degree of flexibility or tailoring of the intervention permitted. Other definitions also emphasize other aspects including the degree of flexibility and non-standardization which complex interventions are subject to; as explained by Hawe et al.2 (2004), complex interventions are non-standard, having different forms in different contexts, while still conforming to specific, theory driven processes. There are many other definitions of complex interventions. These tend frequently to emphasize that they have multiple interacting components, and non-linear causal pathways. Complex interventions are often contrasted in the health literature with ‘simple’ interventions, in particular medical interventions, which are generally seen as having simple linear pathways linking the intervention and its outcome. Such definitions are important because they have direct implications for the sorts of evaluative and other research we do to investigate complex interventions. If researchers see ‘simplicity’ in an intervention, then they may be more likely to argue that randomized controlled trials (as opposed to other sorts of research) are feasible and appropriate. However if they see complexity (non-linear pathways, multiple synergistic components, feedback loops and so on) as the key features, then by implication other types of research may be necessary for illuminating those complex processes. Underlying most definitions is the assumption that ‘simplicity’ and ‘complexity’ are inherent characteristics of interventions. However, there is another possibility: that in fact there are no ‘simple’ or ‘complex’ interventions, and that simplicity and complexity are instead pragmatic perspectives adopted by researchers to help describe and understand the interventions in question. After all, on close examination most interventions, including medical interventions, reveal themselves to be complex. Despite this, it is often helpful to treat them as simple, in order to answer appropriately simple questions. Many interventions may be equally open to either simple or complex forms of analysis. Take a complex public health intervention as an example—urban regeneration programmes. These programmes of structural and social change in urban environments have been often used to improve health and reduce health inequalities via the socio-economic determinants of health. They usually involve complex packages of ‘components’, such as employment, education, income, crime and housing interventions. Research conducted from a complex perspective might consider how and whether these components work individually and together. It might consider the synergies between them, phase changes and feedback loops, and the interactions between multiple health and non-health outcomes, as well as the process by which these components bring about change in communities. However, such research could also adopt a simpler perspective, by ‘unpacking’ the complex intervention into its component parts, and examining the effects of single ‘components’, and perhaps focussing on questions of effectiveness alone. Complexity can therefore be simplified for the purpose of assessing outcomes. A simpler approach might involve simply asking whether investment in urban regeneration as a whole ‘package’ is associated with improved health, or not, and many evaluations have been done this way.3 While the intervention remains the same, the researcher’s perspective on the intervention is different, and the research methods that she/he will adopt will accordingly differ.4 Thus, simpler and more complex perspectives on the same question will yield different, and probably complementary, answers. A simpler perspective may focus on individual-level outcomes alone, whereas a more complex perspective may focus on outcomes at different levels—not just at the level of the individual, but also at the family, and community-level. These different analyses may be more or less useful to different types of user. Some users may want to know about outcomes; some are more interested in processes; many want information on both aspects. Some researchers and users of research require simpler answers, while some want more complex explanations. The world therefore does not after all divide up easily into ‘simple’ or ‘complex’ interventions. Instead there are simple and complex exp