Establishing Evidence Criteria for Implementation Strategies in the US: a Delphi Study for HIV Services

There are no criteria specifically for evaluating the quality of implementation research and recommending implementation strategies likely to have impact to practitioners. We describe the development and application of the Best Practices Tool, a set of criteria to evaluate the evidence supporting HIV-specific implementation strategies. We developed the Best Practices Tool from 2022–2023 in three phases. (1) We developed a draft tool and criteria based on a literature review and key informant interviews. We purposively selected and recruited by email interview participants representing a mix of expertise in HIV service delivery, quality improvement, and implementation science. (2) The tool was then informed and revised through two e-Delphi rounds using a survey delivered online through Qualtrics. The first and second round Delphi surveys consisted of 71 and 52 open and close-ended questions, respectively, asking participants to evaluate, confirm, and make suggestions on different aspects of the rubric. After each survey round, data were analyzed and synthesized as appropriate; and the tool and criteria were revised. (3) We then applied the tool to a set of research studies assessing implementation strategies designed to promote the adoption and uptake of evidence-based HIV interventions to assess reliable application of the tool and criteria. Our initial literature review yielded existing tools for evaluating intervention-level evidence. For a strategy-level tool, additions emerged from interviews, for example, a need to consider the context and specification of strategies. Revisions were made after both Delphi rounds resulting in the confirmation of five evaluation domains – research design, implementation outcomes, limitations and rigor, strategy specification, and equity – and four evidence levels – best, promising, more evidence needed, and harmful. For most domains, criteria were specified at each evidence level. After an initial pilot round to develop an application process and provide training, we achieved 98