Background: Current healthcare evidence relies on relatively narrow efficacy data to make decisions about program impact. This paper illustrates the application of impact indices derived from the RE-AIM (reach, effectiveness, adoption, implementation, and maintenance) framework that takes a broader perspective and includes issues important to decision makers, such as reach, adoption, and cost. Methods: Composite RE-AIM indices that summarize impact and cost efficiency at the individual participant and setting levels are used to compare two different diabetes self-management support approaches. One study, the Diabetes Priority (DP) program, involved 886 diabetes patients from 30 primary care offices, and relied on usual clinical staff for program implementation. The other study, Diabetes Health Connection (DHC), involved 335 diabetes patients in both HMO and fee-for-service settings, and used health education staff. Results: The DP performed better on the setting-level impact index, but the programs produced similar results on individual-level impact. The DP had a greater reach (50% vs 38%); was more effective at the initial follow-up (median effect size [ES]=0.23 vs 0.17); and had greater impact consistency across various populations. The DHC performed better oil several indices, including higher physician office adoption (20% vs 6%) and staff adoption (79% vs 70%), and there was less variability among intervention staff on protocol implementation (median ES=0.0 vs 0.50). Conclusions: Greater use of indices focused on public health and external validity criteria could help identify programs most likely to have a meaningful impact on population health and to fit local settings and priorities.
