Numerous studies, both in mental health and in the broader health literature, document a major gap between our science base and how we deliver care in real-world settings. Helpful innovations often take years to diffuse into routine practice. Ineffective and even harmful practices often persist. Why is this the case? And what can we do about it?

In their thoughtful and well-written article in this issue, Sullivan and her coauthors (1) propose that if we are ever to close the gap between research and clinical practice, we must first close the gap between researchers and clinical practitioners. The authors describe a Department of Veterans Affairs (VA)-based program in which clinicians design and implement mental health interventions with the support of research experts. The interventions are evaluated not only on the basis of clinical and service use outcomes but also on the degree to which activities are controlled by participants rather than driven by experts. Rather than promoting evidence-based medicine, this program can be thought of as developing "medicine-based evidence" (2) to guide clinical practice.

This model is part of a broader class of participatory research that seeks to involve users of evidence throughout the research process. Participatory methods have demonstrated promise in improving uptake and sustainability of clinical and public health initiatives (3), and there is a growing interest in incorporating these approaches in designing and implementing mental health interventions (4). Reflecting its public health orientation, most participatory research has involved partnering with consumers and community agencies. In contrast, the project described by Sullivan and associates works within a health system and engages frontline clinicians and administrators.

Participatory research highlights important questions about the definition, production, and uses of evidence. The fundamental tensions raised by these methods—rigor versus relevance, top-down expertise versus bottom-up practicality, and fidelity versus flexibility—are relevant to all who produce and use evidence in mental health care.

First and foremost, what is evidence? The traditional research paradigm values information on the basis of its internal validity, or freedom from the potential for bias and inferential error (5). However, patients, providers, and policy makers largely understand evidence in terms of its utility. As defined in the American Heritage Dictionary, evidence is a "thing or things helpful in forming a conclusion or judgment." Users of evidence typically must make decisions in the face of high uncertainty, extreme time constraints, and competing demands. Under these imperfect conditions, they seek evidence—perfect or otherwise—to help them make the best choices.

Second, who should produce evidence? Researchers devote much of their careers to uncovering, interpreting, and reporting data. But data become evidence only after they have been used to make decisions. Engaging end users in designing interventions can foster buy-in and sustainability for those treatments. Involving them in selecting relevant outcome measures can help ensure that they act on the results.

Third, how should evidence be used? We are taught that we should bring evidence-based interventions to clinical settings with as few changes as possible. However, there is great heterogeneity across patients, across providers, across organizational structures and hierarchies, and across local financing and regulatory environments. Successful diffusion of innovations typically requires "reinvention" to fit these varying local conditions (6). When we researchers develop an intervention we must both expect and encourage users to tailor it to their needs and capacities.

Bridging "evidence-based medicine" and "medicine-based evidence" will require bridging researchers' and practitioners' attitudes toward evidence and toward each other. Researchers and practitioners have much to learn from each other. If we work together, there is much we can accomplish.

Dr. Druss, a member of the editorial board of Psychiatric Services, holds the Rosalynn Carter Chair in Mental Health at Emory University School of Medicine in Atlanta. Send correspondence to him at 1518 Clifton Road, Atlanta, Georgia 30322 (e-mail, bdruss@emory.edu).