About

Adaptations are a reality when implementing interventions. Adapting interventions to new contexts, settings, and populations is crucial to improving intervention-context fit.

The Model for Adaptation Design and Impact (MADI) provides a lens to help implementation researchers and practitioners be more systematic and intentional about how they are designing or evaluating adaptations to improve the chances that adaptations will have positive impacts. MADI can be used prospectively (as teams are designing adaptations) or retrospectively (as teams are monitoring, evaluating, or conducting research on the impact of adapted interventions).

Figure 1 pic

MADI helps implementation scientists think through some important considerations around adaptations, specifically:

  • Unintended and intended impacts of adaptations: also known as “ripple effects” of adaptations. Although an adaptation may be designed to have one impact (e.g., improve feasibility of implementing the intervention), that same adaptation may have negative, unintended impacts on other outcomes (e.g., compromise fidelity or increase cost).
  • Potential mediators and moderators: most existing adaptation frameworks outline attributes of adaptations (e.g., who made the adaptation, when was the adaptation made, what was the goal of the adaptation), but don’t outline potential pathways for how adaptations are working to impact outcomes, and which outcomes they are impacting. MADI outlines potential mediators and moderators that can help explain why and how adaptations are working to impact outcomes.

MADI also includes two decision-aids to help implementation scientists apply MADI to their work, either as they are designing adaptations or assessing their impacts through research and evaluation.

Decision Aid 1: Prospective Use of MADI

fig 2 pic

Decision Aid 2: Retrospective Use of MADI

fig 3 pic

For more details on applying MADI in your work, see the “MADI Application and Discussion Guide” on the “Resources” tab and check out our publication (Kirk et al., 2020) in Implementation Science.

%d bloggers like this: