Is implementation science a science? Not yet.

Abstract

Getting the science right for implementation is critical for making the processes for improving outcomes more predictable and effective in global public health. Unfortunately, "implementation science" has become a catchphrase for ideas, assumptions, and findings concerning the science to service gap and how to close it. The purpose of this paper is to explore the dimensions of a "science of implementation" that meets the definitions of a science and is focused on implementation variables (i.e., purposeful processes to put innovations into effect so that intended benefits can be realized). A science of implementation is important for accomplishing the goals related to improving the health and well-being of populations around the world. Much of public health involves interaction-based interventions. In a typology of science, interaction-based interventions are created by specifying the nature of certain exchanges between and among individual people or groups. The complexity of developing interaction-based independent variables requires meeting benchmarks for fidelity to assure the presence and strength of implementation independent variables. The paper presents information related to the following tenets: (1) A science of implementation is based on if-then predictions. Science is cumulative. As predictions are made, tested, and elaborated, the facts accumulate to form the knowledge base for science and practice. (2) Implementation variables are interaction-based inventions and, therefore, must be created and established so the specific set of activities related to implementation can be studied. (3) A science of implementation is based on theory that organizes facts, leads to testable predictions, and is modified or discarded based on outcomes. (4) A science of interaction-based implementation depends on frequent measures of independent and dependent variables specific to implementation methods and outcomes. Two examples illustrate the implications for theory, research, and practice. The paper advocates a paradigm shift to a new mental model that values fidelity over tailoring, has one size fits all as a goal, and is concerned with the function of evidence rather than the form of evidence based on RCTs. Global health fundamentally requires scaling implementation capacity so that effective innovations can be used as intended and with good effect to achieve population benefits.