Table 3.
Comparison of some key characteristics of implementation science and complexity science and their integration
| Features | Implementation science | Complexity science | Complexity science and implementation science |
|---|---|---|---|
| Task | The task is specific: getting evidence into clinical practice in an understandable way | The task is context dependent; properties of complexity apply to biology, ecology, physics, computer science, human social systems | Tailored solutions and iterative processes |
| Theoretical assumptions | Heterogeneous and diverse – numerous theories, frameworks, and models | Homogenous – core assumptions of complexity science are characterized by ‘universality’ (i.e., they apply across all complex systems) | Different theories, frameworks, and models require an understanding of complexity features such as unpredictability, uncertainty, emergence, interconnection |
| The intervention | To be standardized to permit generalizability | To be adapted to meet needs | Factoring in complex interventions and complex settings |
| The context | Full of confounders, a ‘problem’ to be solved for successful implementation | An intrinsic part of a complex system; a dynamic environment that must be factored in for any intervention to be successfully taken up | For improvement to be realized, the context must be re-etched or re-inscribed such that its culture, politics, and characteristics are altered |
| Historical underpinnings | Evidence-based practice movement, statistics, and the scientific method | Systems theory, chaos theory; emanating from diverse scientific disciplines | More sophisticated change models can be encouraged to arise over time |
| Aims within health services research | - Describing or guiding the process of translating research into practice (process models) - Understanding or explaining what influences implementation outcomes (determinant frameworks, classic theories, implementation theories) - Evaluating implementation (evaluation frameworks) |
- Description of complex system • Understanding context • Relationships among agents • Dynamics • How rules and governance structures emerge, i.e., self-organization - For prediction rather than implementation |
- Ensure that turning evidence into practice is accomplished without too many unintended negative consequences; improvement might be sustained, potentially through the adaptation of the intervention to different settings - Implementation is not merely based on effective planning but anticipation of a range of possible outcomes |
| Tools and methods | Randomized controlled trials, behavior change interventions, step-wedge designs | Causal loop diagrams, system dynamics modelling, network articulations | Realist evaluation, long-term case study, participatory research, stakeholder analysis, systems mapping, social network analysis |