Use of implementation logic models in the Quadruple Aim QUERI: conceptualization and evolution

Implementation strategies are essential to deliver evidence-based programs that align with local context, resources, priorities, and preferences. However, it is not always clear how specific strategies are selected (vs. others) and strategies are not always operationalized clearly, distinctly, and dynamically. Implementation logic models provide one useful way to conceptualize the role and selection of implementation strategies, plan evaluation of their intended impacts on implementation and effectiveness outcomes, and to communicate key aspects of a project.

This paper describes our initial plans, experiences, and lessons learned from applying implementation logic models in the Quadruple Aim Quality Enhancement Research Initiative (QUERI) a large multi-study program funded by the Veterans Health Administration (VA). We began with two primary implementation strategies based on our earlier work (i.e., Iterative RE-AIM and Relational Facilitation) that were applied across three different health outcomes studies.

Our implementation strategies evolved over time, and new strategies were added. This evolution and reasons for changes are summarized and illustrated with the resulting logic models, both for the overall Quadruple Aim QUERI and the three specific projects. We found that implementation strategies are often not discrete, and their delivery and adaptation is dynamic and should be guided by emerging data and evolving context. Review of logic models across projects was an efficient and useful approach for understanding similarities and differences across projects.

Implementation logic models are helpful for clarifying key objectives and issues for both study teams and implementation partners. There are challenges in logic model construction and presentation when multiple strategies are employed, and when strategies change over time. We recommend presentation of both original and periodically updated project models and provide recommendations for future use of implementation logic models.