Is there "nothing more practical than a good theory" in improving behavioral nutrition and physical activity interventions? And which theories are indeed good enough to help us improve the practice of encouraging people to adopt healthier diets and physical activity patterns? The International Journal of Behavioral Nutrition and Physical Activity (IJBNPA) recognized the importance of this issue and encouraged a 'theory debate' [1
]. Jeffery started the debate by sharing his experiences with and views on applying theories in weight management and weight loss interventions [2
]. His conclusion is that we focussed too much on social cognition models and that these models proved to be not very practical. He was not able to find much evidence from his own studies that using such theories improved the effectiveness of interventions. Rothman contributed to the debate by positing that theory should evolve based on rigorous empirical evidence and that intervention research is one of the best ways to evaluate and refine behavior change theory [3
]. Rothman further stated that already much attention has been given to explaining how theories should be applied, and that now greater emphasis should be given to further refining or rejecting theoretical principles. In the present contribution to the theory debate, we argue that it is still very necessary to further improve the process that guides which theories are applied in behavior change interventions, how these theories are applied, as well as to further improve and integrate existing theories.
What is theory and why do we need it?
Since the publication of Green and Kreuter's Precede and Precede-Proceed models [4
], the health behavior promotion area has recognized the importance of careful theory-based intervention planning. According to these, and other similar planning models [5
], the first step in health-promotion planning is the identification of health problems that are serious and/or prevalent enough to justify spending time, money and other resources. In the second step, the behavioral risk factors for the health problems need to be identified. Step 3 is to investigate the mediators or determinants of these risk behaviors after which these determinants should be translated into intervention goals, change strategies and methods, that need to be integrated in a comprehensive intervention package (step 4) that can be implemented and disseminated (step 5). Each step should preferably be evidence-based (see Figure ).
A Model for Planned Health Education and Promotion
Behavioral theories are mostly used for step 3 of the planning process. Since free choice and autonomy are important values in many societies, and what we eat or how much we exercise are believed to be part of free choice, people choose to a large extend what they eat, and if they are physically active. With very few exceptions (small children, certain groups of institutionalized people), diet and physical activity (PA) behaviors can thus not be influenced directly; instead, we need to influence people's choices. What people choose to eat or do is influenced by a complex, interrelated set of so-called 'mediators' or 'determinants' of nutrition and PA behaviors, including different cognitions as well as environmental factors such as food availability and accessibility [6
]. Successful behavior change interventions are dependent on the ability to influence these mediators. Rothschild posited that these mediators can be divided in three broad categories: motivation, abilities and opportunities [7
]. Thus, complex combinations of motivations, abilities and opportunities determine diet and PA behaviors. Their relative importance, as well as the underlying beliefs of these determinants, are likely to differ across different populations, as well as between individuals within populations, depending on their personal, social, and environmental circumstances. Furthermore, since these circumstances are liable to change, the most relevant specific determinants may also change over time.
Since the second half of the last century, evidence-based medicine came into fashion. The urgency to base our efforts to improve health or to prolong life on scientific evidence was also transferred to public health and health promotion. In 1998, for example, the World Health Assembly stated that all member states should adopt an evidence-based approach to health promotion [8
]. However, what counts as evidence may be debatable. In clinical practice, the randomized controlled trial (RCT) is the 'gold standard' to obtain evidence. An RCT ensures good internal validity, but may lack external validity especially in evaluation of complex behavior-change interventions [8
]. The effects of a diet and PA change intervention, based on an inventory of the mediators of change in a specific population and tested among that population, may not be the same in another population with different motivations, abilities or opportunities. Does this mean that evidence on significant mediators of change and effective interventions are relevant only for that specific population under those specific circumstances and thus that it is in fact impossible to build a real evidence-base for behavior change interventions? We don't think so: we should use evidence obtained in specific populations, and under specific circumstances to build, refine and improve behavior change theory.
The Collins Cobuild English Language Dictionary provides two meanings for the word 'theory'. According to the first meaning, a theory is "an idea or set of ideas to explain something. It is based on evidence and careful reasoning, but it cannot be completely proved". The second meaning of a theory according to Collins is "an idea about something that is based on a lot of thinking but not on actual knowledge or evidence".
What we can derive from research, research that includes RCTs but also formative and process evaluation, as well as other forms of impact and effectiveness evaluations [8
], are evidence-induced general ideas, about which categories of mediators, and which intervention strategies
, as well as intervention channels
may work for influencing certain mediators of behavior change. In other words: we need to build behavior-change theory. These theories should thus be the weighted, systematic, summarized and carefully interpreted results of what has been found in empirical studies directly or indirectly related to nutrition and PA behavior and behavior change. Therefore, we argue that theory-based – in the first aforementioned meaning of the word theory – health behavior interventions is the equivalent of the evidence-based approach in clinical practice. Therefore, theory-based interventions are the only acceptable way to proceed in promotion of healthy diet and PA habits. This, however, implies that we are highly dependent on the quality of the theories and on how these theories are applied in intervention development and implementation. Behavioral nutrition and PA research is a relatively young scientific discipline. Diet and PA behaviors are complex behavioral categories and evidence-induced behavioral nutrition and PA theory is still in its developmental phase and currently too many nutrition and PA behavioral interventions are still based on theory in its second meaning.
In the remainder of this position paper we will, therefore, argue that in healthy nutrition and PA behavior promotion
• we may not always use the available theories in the right way,
• many theories still lack a strong empirical foundation,
• we tend to use theories that are too much focused on the individual and on motivational processes, and
• we may be too inclined to apply a single theory approach.
The Intervention Mapping protocol introduced by Bartholomew and colleagues [10
] suggests specific steps that guide problem-driven development, application and integration of nutrition and PA behavior-change theories. IM proposes a systematic way to proceed from knowledge about behavioral determinants to specific change goals, and subsequently to intervention methods and strategies based on the production of intervention matrixes. Such matrices finally develop into an 'intervention map' that make the translation of objectives to change strategies to actual intervention activities explicit [10
]. In the next discussion paragraph we will therefore refer to approaches suggested in IM that may help to improve the application of behavior theory in intervention development.