The onus is on those producing the evidence to actively engage governments, stakeholders and policymakers, and outline the human and economic advantages of preventive strategies like behavioral interventions over a treatment-focused model of healthcare provision. Related to this, behavioral scientists need to better demonstrate how theory-based behavioral interventions that work in lab and field experiments, and have been shown to be effective in larger randomized controlled trials and in real world contexts, can be implemented in practice. Such evidence should be the focus of evidence presentations to government and policymakers advocating investment in, and implementation of, behavioral interventions. The expanding discipline of implementation science focuses on translation of research findings into evidence-based practice, and is receiving increased attention in the fields of behavioral science, public health, health promotion, and health policy . In the context of behavioral interventions, implementation science examines the pathways and strategies necessary for the uptake and implementation of interventions by policymakers and providers. Evidence on how behavioral interventions can be developed by key workers within existing networks, who will ultimately be responsible for implementing the intervention , and how users of the intervention can be involved in the implementation, is important to ensure that interventions are practically relevant and sensitive to the contextual and cultural characteristics of target populations. In addition, equipment for weed growing research on how theory-based behavioral interventions can be upscaled so their reach within target populations is maximized and the changes in health behavior and health outcomes promised by formative research realized.

Research is needed to identify the conditions necessary to up-scale behavioral interventions in real world contexts, including identifying the partnerships needed to fund, implement, monitor, and maintain interventions; engaging stakeholders to assess the feasibility and acceptability of implementing the intervention in the target community or setting; assisting governmental agencies in developing multi-level and multi-sectorial plans to implement interventions; and developing ways to embed interventions in existing networks throughout development from inception to implementation.In conclusion, interventions based on behavioral theory have been shown to be effective in changing health behavior. However, there is still need for more research on interventions that systematically and precisely map intervention content with theoretical determinants, and the need for greater transparency in the reporting of intervention content and protocols. Arguments that such behavioral interventions do not work in the real world based on observations that pandemics of non-communicable disease continue to rise, and large scale interventions have not shifted population-level participation in health behavior, as my colleague contends, are specious and miss the point. The issue is not that interventions based on behavior theory do not work in changing behavior in ‘real world’ contexts, they do, rather, it is a lack of investment in, and inadequate upscaling and implementation of, these interventions that has failed to translate their efficacy into sustained, long-term change at the population level.Over 50 years positive population behaviors or health outcomes for nutrition and physical activity have fallen or flatlined globally, and in individual countries. Data shows: rising global obesity since 1975 , and inindividual countries including England, Chile, and Australia; falling or flatlining fruit and vegetable consumption in USA since 1994, and in Japan and Brazil since 1965; and rising physical inactivity globally since 2001, in Spain since 1995, in USA since 1997, and in China since 1989.

For health outcomes, European Environment Agency data show loss of healthy life years attributable to non-communicable diseases has grown by more than 20% since 1990. These data illustrate global trends, and their replication in individual countries. Something isn’t working! Noting disjoint between a body of behavioral theory literature that appears to show promise at the individual level, and global and national data that shows no change in population behaviors and health outcomes for half a century, Hallal et al. argue “after more than 60 years of scientific research… more of the same will not be enough” . It appears behavioral theory has a case to answer, and some fundamental questions to face. But is the problem scale-up of behavioral theory in population level interventions and policies, is it intervention designs that act as the vehicle for behavioral theory, or is it simply that behavioral theory itself does not work in the real world?For decades fields such as exercise physiology, public health, epidemiology and the behavioral sciences have undertaken research showing that if behavioral theory is deployed “under scientifically controlled circumstances, behavior change is achievable for increasing physical activity” . However, many “so-called effective physical activity interventions” are small-scale, controlled efficacy trials that do not demonstrate effectiveness or ecological validity, and leave gaps in the chain of evidence between participants, theory, behavior and health outcomes . An intervention is efficacious if it works in cohorts who receive it, whereas it is effective if it works in cohorts who have been offered it. This is confused in the literature, and interventions based on behavioral theory claim effectiveness when available evidence demonstrates only their efficacy. Many trials of interventions based on behavioral theory do not venture beyond controlled environments of phase I-III trials, which seek to establish, respectively, concept, efficacy and comparative efficacy. Thus, at best, evidence demonstrates that impact on those who receive the intervention exceeds impact on those who receive alternative interventions. But still, this shows only that an intervention is comparatively efficacious for those who receive it, not that it is effective, or comparatively effective, in cohorts that are offered it.

The problem is this: the features of design and implementation associated with good phase I-III trials to establish concept, efficacy, and comparative efficacy, have important limitations for informing practice and policy decisions, which require more generalizable information relating to outcomes of societal consequence, such as a sustained impact on health outcomes at population level. Such impact, or the potential for it, must relate to real world effectiveness “as evaluated in an observational, non-interventional trial in a naturalistic setting” . To establish effectiveness, phase IV trials require a more diverse set of methods than those required to establish concept, efficacy and comparative efficacy in phase I-III trials, and must involve a diversity of settings, participants and deliverers . However, in reviewing studies purporting to examine effectiveness of physical activity interventions in the real world , Beedie, Mann and Jimenez found that many still tried to adopt laboratory style methods and controls that would be impractical or uneconomic in real-world settings. Some authors have advocated the RE-AIM framework as a Phase IV tool to develop the effectiveness of interventions shown to be efficacious at phases I-III. But, with its focus on ensuring reach, adoption, implementation integrity, and maintenance of the features of the intervention over time, RE-AIM merely attempts to deliver effectiveness by maintaining the controlled environment of phase I-III trials in the real world, which as well as being impractical or uneconomic, is also likely to be futile. Establishing effectiveness in phase IV trials is difficult, and requires longer timescales, and greater scale and resources than establishing concept, efficacy and comparative efficacy in phase I-III trials. As such, it is not surprising that, in an area where research funding is relatively sparse, pipp horticulture and doctoral studies are often the bricks contributing to edifices of knowledge, genuine phase IV effectiveness trials are rare. Nevertheless, there is a moral obligation to conduct them, otherwise advocacy for behavioral theory interventions based only on efficacy evidence risks wasting participants time and taxpayers money on unproven interventions in unproven populations.Analyses of national participation data suggest interventions based on behavioral theory may be recipients of individual behavior change, rather than the stimulus for it. This is because populations’ behaviors are qualitatively different to individual behaviors, and incorporate individual behavioral volatility within their steady state. Forexample, in England two national surveys, Active People and Taking Part, show population participation in sport and related physical activity has flatlined for 10 years, with no sustained change beyond +/− 2%. Furthermore, data synthesis across six surveys shows falling or flatlining participation for 25 years. However, both cross-sectional retrospective report data and panel time-series data from the surveys also shows considerable individual behavioral volatility, with circa 20% of the population dropping out or doing less sport, 20% taking up or doing more sport, 20% maintaining participation, and 40% consistently doing no sport.

Consequently, within any 1 year circa 40% of the population change their sport participation behavior, but aggregate population level participation is unchanged. Thus, steady state population behaviors incorporate considerable individual behavioral change. This suggests behavioral theory interventions are reflecting and facilitating individual behavior changes that take place as part of the steady state behaviors of populations, with participants often presenting as already motivated to change [88, 93]. Sport England’s Get Active: Get Healthy first-year pilots, for example, claimed to be the stimulus for more than 30,000 people becoming active, but the evaluation showed the majority of participants were “ready to change” when they joined. This suggests the interventions were the recipients rather than the stimulus for individual behavioral changes, which are to be expected as a normal part of steady state population behaviors. It is known that poor health outcomes, particularly non-communicable diseases, correlate with social deprivation, low employment, poverty, poor housing, and other indices of multiple deprivation. Behavioral theory provides neither the explanation nor, through interventions targeting individuals, the solution to such problems, which must focus on wider causal systems that underpin the social practice and economy of behaviors such as low physical activity and poor diet. Undoubtedly, it is the focus on the individual rather than the population that undermines the real-world effectiveness of behavioral theory. The etiological model on which it is based – that poor health outcomes are caused by exposure to a substance, for example, sugar, and that health outcomes can be improved by modifying or moderating individual behaviors to remove or reduce exposure – is fundamentally flawed. This is because solutions – interventions based on behavioral theory – have no relationship to causes – the factors that lead to behaviors in the first place. Furthermore, behavioral theory is assumed to be universal: that is, it is assumed the same behavioral theory can address any behavior, be that smoking, alcohol consumption, poor diet, or low physical activity – the transtheoretical model, which was developed for smoking cessation, is a case in point. Cleary these behaviors are underpinned by different antecedents, so why would we assume they can all be addressed by the same theory? Furthermore, categories of behavior are not homogenous – the existence of health inequalities is, in itself, evidence that the factors that lead to behaviors in relation to, for example, diet, differ across the population, and so poor diet is an agglomeration of behaviors rather than a single behavior. Why would we expect that these multiple complex behaviors could all be addressed by the same theory?I have argued that while interventions based on behavioral theory have been shown to be efficacious in the controlled environments of phase I-III trials, there is no evidence from genuine phase IV effectiveness trials to demonstrate they work in the real world. However, crucially, I argue that evidence from controlled trials of behavior change interventions simply capture individual behavioral volatility that is a normal part of steady state population behaviors. Furthermore, such interventions fail in shifting population behaviors because they focus on individuals rather than on the multiple complex factors that drive the distribution of behaviors in the population. As such, behavioral theory within such interventions is not an active ingredient, rather it is a dormant recipient of behavior change. Put simply, behavioral theory has no active influence on changing behaviors in the real world.I am grateful to my colleague for raising important points on the implementation of theory-based behavioral interventions and the need for more evidence for the effectiveness of behavioral interventions in ‘phase IV’ trials. These are good points that have been made many times elsewhere, including my opening statement. However, as an argument against the proposal, his statement is not fit-for-purpose. As I predicted, my colleague claims that interventions based on behavioral theory do not work in changing behavior in ‘real world’ contexts because there has been no year-on-year change in rates of non-communicable diseases and health-related behavior participation at the population level. He also suggests that behavior theory focuses solely on individual behavior, targets only the motivated, and fails to incorporate structural determinants of behavior. Here I illustrate how his arguments reflect a poor understanding of behavioral theory, and are not based on appropriate evidence, or, in some instances, any evidence at all.