Fifteen years ago, I reviewed the central problems in the field of mental health services research and identified unmet needs for behavioral health services, lack of adherence to pharmacological therapies, inadequate retention in behavioral health care, and poor resource allocation of mental health providers as persistent problems requiring solutions.  I found that, although there was a considerable knowledge base available to solve these problems and a substantial body of recommendations, we had no Randy Moss (an American football wide-receiver for the New England Patriots) at the other end to grab the recommendations and run with them.

So, I asked myself, why?  Why is there such an enormous gap between the recommendations about how to solve enduring problems and the implementation of these recommendations? In fact, Lavis et al. (2003) referred to this gap as “the paradox of health services research, i.e., if it is not used, why do we produce so much of it?” The answer to this question warrants serious reflection and debate among researches, policy makers, funders, and the wider public who support our field with their tax dollars.

Some people believe that health services problems do not get solved because the nature of the evidence is inadequate and have called for what are referred to as “practical clinical trials” (PCTs).  In an article published in the Journal of the American Medical Association, Tunis, Styer and Clancy (2003) stated: “The widespread gaps in evidence-based knowledge suggest that fundamental flaws exist in the production of scientific evidence, in part because there is no consistent effort to conduct clinical trials designed to meet the needs of decision makers.” The numbers of PCTs are restricted chiefly because the key funders of clinical research, such as the National Institutes of Health, do not focus on supporting such trials.

Others have suggested that the knowledge base is inadequate because research evidence by itself may be insufficient.  For example, Claxton, Cohen and Neumann (2005) argue that evidence from randomized controlled trials (RCTs) does not tell us whether an intervention should be adopted. RCTs are not intended to address implementation issues or consideration of the benefits relative to the costs involved in the adoption of the intervention, or to provide contrasts to other alternative options of treatment.

Some may view the problem of not having adequate evidence-based information as due to a lack of dissemination of evidence-based research and offer a short-sighted view of the need to invest in the process of implementation. The distance between the people who produce the information and the people who use the information is vast.  We do not do much to minimize this distance--not even asking if the information we are producing will be valuable to the end-users. My message in this article is simple: We now need to move health services research from a science of recommendation to a science of implementation. We cannot wait any longer for others to implement our recommendations.

One likely problem with taking off-the-shelf recommendations from health services research is that they are filled with assumptions not supported by observable deliberations in real-life circumstances. This disconnect with everyday practice suggests that to move from a science of recommendation to a science of implementation, we need to better understand the assumptions involved in using the evidence for decision-making or for anticipating the actual behaviors of consumers, payers, providers or policymakers. 

For instance, what happens if you assume that providers will gravitate toward the use of evidence-based practices because they mainly focus on the simple question, “What is the best evidence-based approach for patients with selected conditions?”  This question will prove meaningless if based on flawed assumptions, e.g.,  that there are practice guidelines accessible for given conditions, and that the physician treating the patient is familiar with and informed by the guidelines and has the infrastructure support to implement them. Unfortunately, we know many conditions do not have strong evidence-based guidelines (e.g., eating disorders, peptic ulcers, chronic pain), and dissemination of guideline - concordant care is in its infancy.

The number of evidence-based research findings in both the natural and social sciences that have been translated into clinical practice is discouragingly small. A recent study in Canada assessing why decision-makers fail to use research findings found that 76% of those surveyed commented on the lack of relevant usable research for knowledge translation and integration. Practitioners, managers, and policy-makers habitually disregard pertinent research evidence on the grounds that inferences drawn from randomized control trials are too narrow and do not represent their circumstances).

The focus in research design has correspondingly shifted from efficacy and internal validity toward effectiveness. As a consequence, we have incorporated research designs that account for the contextual factors that influence a particular intervention in practice or “real-world” settings. In addition, there has been great progress in identifying, critically appraising and synthesizing the published evidence.

Thus, the shift is towards identifying what are workable and pragmatic approaches to improve care. At the same time, those conducting evidence-based practice have begun to recognize the value of integrating patient views and understanding how they engage in decision-making as a way to bridge research and practice.  We have also started to acknowledge that successful translation of evidence-based health services requires evaluating the evidence of the impact of the services, resources necessary to provide them, context where services will be implemented, mechanisms for change in the organization, and preferences of consumers relating to acceptance of barriers to care.

However, the progress, while notable, is not enough.  We need to move to a new frontier, one focusing on knowledge of implementation and facilitation, where the spotlight is not only on the nature of the evidence but on the context of uptake and on how to facilitate uptake in that context. For example, when we start thinking about to whom research knowledge should be transferred, we also have to consider that knowledge transfer must be fine-tuned to the types of decisions that providers, payers, and consumers face and the types of contexts in which they live and work.

It is critical that we engage with policy makers, communities, and healthcare providers if we truly want to disseminate the research evidence. Effectively implemented interventions are those where there is repeated interaction between the clinician and an expert who has been trained in the principles of, for example, academic detailing, or frequent interaction between the clinician and someone to whom he or she routinely turns for guidance. Making this transition to implementing the evidence goes beyond merely having the information; it requires having a support person who will facilitate uptake and ensure that the translation and implementation of evidence remains a priority.

Horton’s work (2006) suggests what it might take to understand and fix healthcare policy so it can work. Her ethnographic study of Latino mental health clinics in the northwestern United States shows that the new private sector measures of productivity take a toll on both the Latina clinicians, whose invisible work subsidizes the system, and on the particular categories of patients-the uninsured and immigrants with serious psychosocial issues. She took the time to understand what it means to be a productive clinician when working in the context of the requirements of the health care system (e.g., managed care, cost containment).

Having a policymaker read her work could open a window on how policymaking can affect practice in real-world settings, to create solutions that would remediate these harmful consequences. For example, she found that high demand for system accountability and worker efficiency encouraged providers to take shortcuts by treating individuals as mass categories. In particular, she observed that, while clinicians attempted to buffer their patients from the impact of such reforms, they also resorted to means to increase their productivity such as by dropping repeated no-show patients and denying care to the uninsured.

As we reflect on how to move from a science of recommendation to a science of implementation, we must consider the importance of engaging with communities and clinicians, how to become the “Randy Moss receiver” for public facilities, to be able to anticipate the negative consequences that our policies may have on our “uptakers.” And herein lies the importance of community and practical clinical trials.  If we create an operational infrastructure to conduct implementation research within a community or a public clinic, we not only transport ourselves to the scenarios and understand the circumstances of a given community, but we actually conduct implementation research to make an impact in the next 25 years.

The expected and necessary changes that will impact and restructure healthcare will require research evidence to be used for significant policy and system change. With this research, serious attention will need to be devoted to the translation and implementation of knowledge.  It is here where all of us, as the responsible generation of health services researchers, and all of the younger investigators in this area in particular, must make sure that we provide a science of implementation that is not only inspirational to practitioners but also pragmatic enough that it will result in ‘uptake’. In this way, we may be able to ensure that 25 years from now we are not reviewing the same old problems.

This article was based on the following publication: Margarita Alegria (2009) Academy Health 25th annual research meeting chair address: From a science of recommendation to a science of Implementation. Health Services Research, 44(1). doi: 10.1111/j.1475-6773.2008.00936.x The author wishes to acknowledge the participation of the Harvard Medical School Center for Multicultural Mental Health Research.