Current issues and new directions in Psychology and Health: Theory, translation, and evidence-based practice

Abstract

Health psychology has at least two masters; cognitive behavioral theory and clinical and public health practice. We are expected to contribute to theory by creating new models and adding to existent models of human behavior, and to contribute to improvements in health outcomes for the public. In this brief note, we propose that translating concepts from practice into theory to create interventions that meet the standards of evidence-based practice will satisfy both masters. However, the integration will require a re-examination of our current understanding of how to use theory, the process of translation, and the development of pragmatic evidence-based practice.

Theory and health research

We propose a reorientation to the role of theory for health research to make maximal use of the specific powers of each theory for solving health-related problems. To do this, we see the need for two types of theoretical models: descriptive models and process models. Descriptive, epidemiological models identify problems and set the focus for process analyses by characterizing the distribution across segments of the population of specific diseases or of patterns of underuse of appropriate screening, prevention, or illness-management behaviors. Models to explain how things happen–process models–require the development of content variables at different levels of analysis. Content specific variables will be required to operationalize the manner in which people conceptualize specific health threats and specific behaviors for threat control, that is, what and how they think about particular diseases and about corresponding screening, prevention, or management techniques. Models of health threats and procedures for prevention and treatment are nested in the individual's experiential and conceptual framework of the self (e.g., prior experience with acute and chronic illness and efficacy in behavioral management), and nested in specific social contexts (e.g., observations and communications about the effects of illness as well as the use and effects of treatment on similar and dissimilar others) (Rothman, 2000; Schwarzer & Fuchs, 1996). A complete social-cognitive behavioral model will describe how contextual factors communicate with and regulate the processes involved in self-management behaviors in terms of both descriptive and process-based information.

Our approach contrasts with that of many investigators who have undertaken comparisons among theories of health behavior in the belief this will produce the best model for understanding health behaviors. Given the lack of specificity of behavioral theories, our belief is that they are best treated as complementary views of behavioral processes rather than as competitors; it is more useful to determine how each can be used most effectively and how they are related to one another. For example, a theoretical concept may be useful for predicting a health outcome without elaborating and testing the mechanisms underlying the relationships. Consider individual differences in hostility or conscientiousness, which identify respective individuals who are likely to develop cardiovascular disease and individuals likely to be adherent to treatment, but do not explain (1) how individuals differing on these factors perceive and represent illness threats and treatment options or (2) how the impact of efforts to control symptoms or information from objective indicators (e.g., blood sugar levels) affect ongoing performance. The descriptive data however, suggest hypotheses and focus research on the processes underlying the relationships. The process can be behavioral, such as healthy and risky behaviors or ways of organizing and coping with daily life stresses, and they can be physiological. The unique contributions of descriptive and process models are complementary rather than in competition.

There are many ways in which comparisons among theories can create barriers to understanding the processes underlying behavior and minimize their usefulness to practice. For example, when two theoretical models using different measuring instruments are compared, one cannot be sure whether the empirical advantage of one is due to its conceptual structure, its methods, or deficits in the concepts or methods of the model that faired less well. Failing to recognize that not only concepts but also methods are involved in theory comparisons is the source of much argument about the relative merit of social cognitive behavioral models and models based on attitude measurement, such as the theory of planned behavior.

Conclusions about the relative “power” of the theories in accounting for variance in health behavior also arise in many cases from the inclusion of intention in the planned behavioral array and its omission from the cognitive behavioral model. It has long been argued that variance accounted for by intention may not be part of a causal chain (Bentler & Speckart, 1979), and that intention instead may be a noncausal predictor or proxy for the behavioral outcome. Identifying predictors is only a means of arriving at explanation; it does not constitute explanation (Weinstein, 2007). It is also unclear why intention is excluded from cognitive behavioral models; comparisons of private versus public statements of commitment have been examined by cognitive theorists as causal antecedents of action in cognitive models. As it is unclear whether investigators are comparing theories or methods and whether the constructs are assigned somewhat arbitrarily to one theory rather than another, one can question whether these comparisons contribute to an understanding of process, i.e., how health behaviors are initiated and maintained. For example, in recent comparisons of theories involving self-regulation, though investigators have used a validated measure of illness representations for self-regulation, they have ignored the output side of the self-regulatory system, i.e., the response (its efficacy, goals (symptoms or blood sugar), time for effect, etc.) (Powers, 1978). The findings derived do not compare theories and the idea that they do so is not helpful for understanding the processes or mechanisms shaping health and illness behaviors in specific contexts.

The value of seeing the complementary nature of theories to understand the processes underlying health behavior can be seen in efforts to compare operant and self-regulation theory. Despite their paradigmatic differences, both theories can be seen as using similar if not identical concepts: eliciting cues (e.g., a light in the operant lab, a symptom in self-regulation), responses (lever press; taking a medication), and reinforcements (food; symptom removal). The differences in terminology reflect the historical and investigative contexts in which each theory was developed. Rather than treating these theories as competitors, it is more useful to introduce reinforcement schedules into self-regulation models, and perceptual and cognitive processes into operant models to create a more powerful model for examining the processes involved in the maintenance and as well as the initiation of changes in health-related behaviors.

Translational research

The revolution in cell biology has intensified calls for translating laboratory science into treatment; a call that asserts a unidirectional view of what should be a bidirectional process. A unidirectional view makes sense when a biological model provides a sufficient explanation for a disease. It is insufficient, however, if environmental and behavioral factors affect exposure and/or alter gene expression. If ecological, cultural, economic, and social variables affect exposure to pathogens and/or gene expression, these factors need to be incorporated in models to account for disease onset and progression (Feinstein, 1974). Health behavior research requires a bidirectional view of translation in which the concepts of bench science are translated to the community and clinic, and in turn, concepts generated in the community and clinic are incorporated into the bench science model to create a more complete science of the disease process. Bidirectionality is critical for the development of process-oriented social-cognitive behavioral models of health behavior. Such models are necessary for learning how people represent health threats and procedures for threat management. How patients make sense of their experience with disease and treatment can be observed in clinic and community settings and their experiences in these settings will affect their representations of diseases, treatments, and themselves and influence their behavior. The concepts generated in people's lived environments are then incorporated in “laboratory” based theoretical models (Leventhal, 1970). Social-cognitive behavioral models of these processes will be the product of a bidirectional translation that generates theories and data that will satisfy the needs for evidence-based practice.

Evidence-based practice

The need for comprehensive, process based, social-cognitive-behavioral theory becomes clear when one considers the construction and testing of behavioral interventions to meet the two criteria for “evidence-based practice.” The first criterion is evidence for efficacy in pilot trials and effectiveness in rigorously designed and conducted multisite clinical trials demonstrating that a treatment can prevent and/or slow disease progression. A treatment, medication, or life-style change must also meet a second criterion to merit clinical application: usability in clinical settings (Evidence-Based Medicine Working Group, 1992). The diabetes prevention trials among high risk individuals provide excellent examples of effectiveness for behavioral interventions (Gillies et al., 2007; Yamaoka & Tango, 2005). In both the Finnish and US trials, participants in the behavioral arm were 58% less likely to become diabetic in comparison to participants in standard treatment (Knowler et al., 2002; Tuomilehto et al., 2001). However, the behavioral interventions likely fall short of the usability criterion as the intensity and complexity of this intervention (16 individual sessions, phone contacts, etc.), in the US trial (Knowler et al., 2002) make it unusable given the reimbursement structure for clinical practice.

Translation, theory, and evidence-based practice

Combining our view of theory and the translation process with both requirements of evidence-based practice defines a research pathway satisfying to both theory and practice. Descriptive/predictive models will point investigators to areas in which the behavior of institutions, groups, and individuals affect health outcomes and dynamic/process oriented models will provide concepts for developing and testing interventions. Understanding the dynamics of the self-management processes for specific diseases and treatments, and observing how these processes are shaped by interpersonal communications and cultural and institutional contexts, are steps toward the development of effective and efficient (usable) interventions. For example, what practitioners say and how they examine a patient's body and what patients hear when they compare their symptoms and response to treatments with other patients, family members, and friends will identify facets of interventions essential for initiating, maintaining, and making effective “self-management” satisfying and eventually, habitual (Rothman, 2000).

To satisfy the masters of theory and the criteria for evidence-based practice, it is necessary to identify the factors that are essential for initiating and sustaining behavior change in specific settings. Theory-based experiments or “efficacy trials” are an essential precursor to effectiveness trials that confirm the effectiveness and usability of behavioral interventions across multiple sites. We agree with the Editors of Psychology and Health that there is:

“… a need for more small-scale experimental studies that can help to identify the active components of an intervention and establish the mechanisms by which they may affect outcomes. Provided that interventions are theory-based, this provides an important means of testing the relationships predicted by our theoretical models and the validity of the models themselves.” (Yardley & Moss-Morris, 2007, p. 2)

Research that describes the characteristics of individuals, illnesses, treatments, and the environments in which these are experienced, combined with studies of process and the need for a bidirectional approach in conceptualizing interventions for clinical settings, will address the call by our Editors and provide the evidence base that can lead to larger scale effectiveness tests of usable interventions. In agreement with prior commentaries (Michie, Rothman, & Sheeran, 2007), we repeat Kurt Lewin's assertion: “[t]here is nothing so practical as a good theory”, and add his prescient support of bidirectional translation: “no action without research, no research without action” (1951).

Acknowledgements

The authors would like to thank Leigh Alison Phillips and Dr. Pablo Mora for their valuable comments on an earlier draft of this editorial. Preparation of this article was supported by grants from the National Institute of Health (R24-AG023958, Center for the Study of Health Beliefs and Behaviors; ROI-AG16750).