Skip to Main Content


All interventions are complex, but some are more complex than others: using iCAT_SR to assess complexity

Graham F Moore, Rhiannon E Evans, Jemma Hawkins, Hannah J Littlecott, Ruth Turley
Editorial Article

The term 'complex intervention' is commonly used by health researchers. However, what makes an intervention complex (or indeed whether there is any such thing as a non-complex intervention), remains contested.[1,2] Early guidance from the UK Medical Research Council (MRC) positioned complexity as an internal property of a new way of working, stating that "complex interventions are those that include several components".[3] Hence, 'complex' multi-component interventions were contrasted with simpler mono-component interventions, such as drug therapies.

More recently, complex systems perspectives have emphasized context as a primary source of intervention complexity.[2,4] From this perspective, an intervention is not a discrete package of components, but a process of changing what complex systems do. Intervening within a complex system simultaneously involves disrupting prior ways of working and introducing new ones. Changes in outcome trajectories may be shaped both by what is displaced and by what is introduced.[5] Rather than an absolute property of new components, intervention complexity can therefore be understood as a relative construct, linked to usual practice within the system, and encompassing challenges associated with disrupting and replacing often entrenched ways of working.[6] Updated MRC definitions responded to these shifting views of complexity to some extent, recognizing the number and difficulty of behaviour changes required for full implementation and the extent to which local tailoring is permitted as key dimensions of complexity.[7] Updated guidance acknowledged that effects of complex interventions vary by context. However, complexity continued to be conceived largely as a property of intervention components.

The intervention Complexity Assessment Tool for Systematic Reviewers (iCAT_SR) builds on the updated MRC definitions of complex interventions, providing systematic reviewers with a structure for locating interventions on a continuum according to a number of dimensions of complexity.[8] As Mark Petticrew argues, most seemingly simple interventions have aspects of complexity on closer inspection, and hence the movement beyond a simple/complex binary is welcome.[1] The six core dimensions of iCAT_SR represent an elaboration of dimensions within the 2008 MRC guidance, including the number of components, the diversity of intended outcomes, the number of organizational levels targeted, and the level of skill required to deliver the intervention. It maintains a traditional complex interventions perspective, with system complexity remaining in the background. Contextual contingency of intervention effects, for example, is treated as a secondary optional item. A focus on implementer skill as the primary dimension of implementation complexity excludes other important aspects of complexity, which may influence whether a new way of working becomes assimilated or washes out of the system.[2] Furthermore, it encourages comparisons between interventions on the basis of their components, rather than in terms of the complexity of their mechanisms of action.

In conjunction with other tools with a comprehensive focus on system complexity, such as the Context and Implementation of Complex Interventions (CICI) framework,[9] iCAT_SR may be helpful in enabling systematic review authors to conceptualize and test how complexity within a new package of components, and within the system, shapes outcomes. The developers of iCAT_SR argue that their tool may be most useful "in reviews where understanding the complexity of the intervention, and understanding the impacts of this complexity on intervention effects, are important parts of the review question".[8] These are important questions in domains such as school health. Simpler interventions based on health education are often ineffective, although more complex multilevel interventions are often derailed by implementation failure.[10,11] Hence, iCAT_SR may help reviewers understand in what circumstances, and in what ways, increasing intervention complexity increases impact or leads to implementation failure. The tool may be particularly useful in the context of mixed-method reviews that consider the totality of evidence for a particular intervention approach, rather than reviews focused purely on effects. Indeed, given the often-limited description of interventions within effectiveness articles, a fuller picture of complexity will likely require engagement with companion publications, such as intervention manuals and process data.

For tools such as iCAT_SR to be useful for systematic reviewers, primary researchers need to report complex interventions, as intended and as delivered, in sufficient detail, as advocated within frameworks such as TIDieR.[12] Attempts to rate the degree of synergy of interactions between components, or subgroup variations in effects, may be hypothetical, unless empirically investigated within primary studies. Disaggregation of effects by socioeconomic status, for example, is necessary within primary studies to inform judgements on whether effects vary by socioeconomic status.[13] The iCAT_SR tool represents a potentially useful supplement to a number of methodological developments in primary research, providing researchers with a structure for conceptualizing the complexity of new intervention components (relative to what already is in place) throughout the intervention development and evaluation process. The MRC has funded the development of new guidance on intervention development and exploratory studies to inform full-scale evaluations,[14,15] while guidance is underway on taking account of context throughout the intervention development-evaluation-implementation process. Complex interventions in complex systems pose an almost infinite number of potential uncertainties, with intervention complexity often inversely associated with the extent to which evidence reduces uncertainty in health decision making.[16] In order to maximize the value of evidence for decision makers, evaluators need to prioritize key aspects of complexity for empirical study.[17] Hence, tools such as ICAT_SR may assist in anticipating and planning for implementation challenges, and designing evaluations that are proportionate to the uncertainties posed by an intervention and that focus upon core elements of complexity. This could help to maximize the role of evaluation in reducing decision-maker uncertainty.