The overall aim of the scoping review was to examine how the “biopsychosocial” has been conceptualized in the existing literature on PA behavior. Overall, we only found a limited number of studies that analyzed PA behavior from a biopsychosocial perspective. Notably, only six of the included 13 articles were empirical studies concerned with PA behavior.
Underlying Theoretical Models or Frameworks
Broadly speaking, the models in the original articles either depicted correlates of PA behavior, or the influence of PA on various biopsychosocial outcomes such as health, quality of life, or successful aging. Several justifications for using a biopsychosocial model or a similar theoretical approach (i.e., transdisciplinary approaches or social-ecological models), which also focuses on biological, psychological, and social aspects, were found in the original articles. Nearly all articles pointed out that a biopsychosocial perspective is fitting to simultaneously examine relationships between biological, psychological, and social dimensions with regard to correlates and effects of PA behavior. Thereby, authors of original articles often referred to the biopsychosocial model’s potential to describe a phenomenon from a multidimensional perspective that takes interactions of various factors into account.
A total of five included articles employed the socio-ecological model in the tradition of Bronfenbrenner . Even though socio-ecological models are models in their own right, we included these articles in the present review as long as it was apparent that biological, psychological, and social factors were part of such a framework; thereby, this work implicitly followed a biopsychosocial perspective. Socio-ecological models in the included articles were concerned with explaining PA behavior or describing a wide range of factors that potentially influence PA behavior in specific populations . Thus, socio-ecological frameworks are used to model not only behavioral influences within and between individuals but also environmental and political influences. Within-person influences often include the biological and psychological components whereas the social and cultural components are summarized under influences between persons or as environmental influences .
In contrast to socio-ecological approaches, articles that referred to the biopsychosocial model in the tradition of Engel were not only concerned with influences on PA behavior but also focused on possible biopsychosocial effects of PA behavior; thus extending the scope of the socio-ecological model through considering PA behavior either as the outcome variable or the independent variable.
Taken together, even though both models are theoretical frameworks in their own right, they share a transdisciplinary perspective on PA behavior. When looking at the broader field of health and illness research, there have even been attempts to combine Engel’s biopsychosocial model with Bronfenbrenner’s ecological model in order to better explain health- and illness-related changes in a person . The present scoping review demonstrates the various ways that PA behavior can be included either in a biopsychosocial model or a socio-ecological framework, with neither one offering a concrete basis for the inclusion of PA.
Variables Included in the Biopsychosocial and Socio-ecological Models
The models that were used in the reviewed articles differed to a large extent with regard to the integrated components and the assessed variables. While some articles followed the original classification into biological, psychological, and social aspects, other articles used different categories such as proximal, distal, interpersonal, intrapersonal, behavioral, environmental, and cultural aspects. These different overarching categories directly go back to the nature of the included models. Whereas most articles that were based on the biopsychosocial model in the tradition of Engel employed the traditional categorization into biological, psychological, and social aspects, socio-ecological models directed attention also to environmental and policy features. Since socio-ecological models in included articles were mainly adopted for studying PA behavior in specific places and populations, a greater focus on the characteristics of places that hinder or facilitate PA behavior can be observed.
Overall, a broad variety of variables was included in the theoretical frameworks. The reasons for such a variety might be traced back to the objectives of the included articles and to the study samples. For example, different aspects may be of significance or of greater significance for the prediction of PA in children and adolescents  than for the effects of PA on the quality of life of former colorectal cancer patients . Even though this observed variety of variables makes a clear comparison of approaches nearly impossible, it also demonstrates the potential of a biopsychosocial perspective to be used in a variety of research contexts and to inform transdisciplinary PA research.
The systematic integration of relationships between the biological, psychological, and social components is emphasized in the original work of Engel [7, 35]. However, only few of the included articles empirically analyzed interactions between the individual components [27, 28, 32]. Complex statistical analyses such as path analyses and bivariate correlations , moderation and mediation analyses [27, 28], structural equation modeling with a focus on testing theoretical relationships between latent constructs , and multivariable linear regression models with subgroup analyses  were performed to analyze such complex interactions of model components.
All in all, the aforementioned points illustrate the challenges that arise when applying a biopsychosocial perspective on complex phenomena such as health and/or PA. These challenges become even more complex when differences between individuals and within individuals over time are taken into account.
Implications for Future Research
The present scoping review shows the benefits of a biopsychosocial perspective on PA behavior. Both the empirical and non-empirical articles in this scoping review advocate for a multidimensional and complex perspective in research on the biopsychosocial correlates and effects of PA behavior. Due to the complex nature of PA behavior, research on correlates and effects of PA should be informed by theoretical approaches that aim to capture this complexity, either with regard to the interactions of multiple correlates of PA behavior or with regard to the complex and interrelated effects of PA.
Even though it seems to be common sense that PA behavior is complex and influenced by a myriad of factors, most empirical evidence is still not based on comprehensive approaches. This point is also demonstrated by the relatively small number of empirical studies identified in this review that explicitly follow a transdisciplinary perspective when examining the correlates of PA behavior.
The included empirical research based on a biopsychosocial perspective convincingly demonstrates that no one factor explains PA behavior; rather it is an interaction of factors at different levels. Grounding empirical research on sound theoretical frameworks, such as the biopsychosocial model or the socio-ecological model, facilitates the identification of important contextual factors and potential confounding and moderating effects when examining influences on PA behavior. However, one major drawback of the models that we analyzed in our scoping review was their missing focus on the complex interactions between the various factors. It is precisely these interactions that add great value compared to a unidimensional view when examining such a complex, multidimensional, and dynamic phenomenon as PA behavior. In future research, approaches are needed that aim to capture these interaction effects.
Thus, for future transdisciplinary PA research, it is a reasonable approach to examine multilevel influences on PA behavior. Such an approach to PA behavior necessitates application of complex statistical procedures (such as growth curve modeling, structural equation modeling, latent class analysis, etc.) that allow identification of intra- and inter-individual and contextual differences as well as dynamic interactions between different factors. Additionally, a multilevel perspective also requires long-term transdisciplinary work from different scientific disciplines such as sports medicine, epigenetics, health psychology, sport sociology, and public health, employing different methodological approaches to capture the full complexity of the underlying phenomenon (i.e., PA behavior). Through considering factors at different levels and particularly their interactions, which necessarily makes such research more complex, it becomes possible to better understand influences on PA behavior. Knowledge gained through such a transdisciplinary perspective will help to develop and implement PA promotion programs, public health strategies, interventions, and policies that more effectively target specific contributing factors for PA behavior in at-risk groups for low PA levels.
Transdisciplinary approaches are also required for recognizing and understanding the effects of PA behavior on an individual and societal level. Asking solely for effects on one level (such as the physiological level) might overlook or even disregard important benefits of PA on other levels (such as the psychological or social) that might also affect health and quality of life, or the outcomes of rehabilitation programs or treatment strategies. Research on the effectiveness of health-related behavioral interventions might also be improved if it considers the biological, psychological, and social benefits of PA.
While transdisciplinary approaches in PA research are more common in some disciplines (e.g., sports psychology and sports medicine or social-psychology), the collaboration between others is still missing. One such transdisciplinary approach might be between the evolving field of epigenetics and sociology and psychology. Outside the field of sport science, researchers have already begun to argue that epigenetics should expand its scope beyond molecular biology research only , and rather complement its research agenda with sociological and psychological models, which might help capture environmental influences on gene expression . By including epigenetics as a sub-aspect of the biological component, future transdisciplinary PA research could generate new insights into the genesis of many health-related phenomena through a focus on interactions between genetics and PA , or the influence of the environment on gene expression in the context of adaptations to training, exercise, and PA .
Limitations of the Scoping Review
Some important limitations of our review need to be considered. The first limitation lies in the small number of included studies. The relatively strict inclusion and exclusion criteria could be a reason for this. However, they were necessary to limit the scope of the review. We acknowledge that further insights could be gained by broadening the search strategy, i.e., also including the areas of exercise and sports.
Additionally, data may be incomplete because some studies might have been published in another language or indexed in other databases. However, through screening of reference sections of eligible studies, we aimed to lower the risk of bias across studies.
Since we solely included the terms biopsychosocial and transdisciplinary in our search, articles that only implicitly employ such a perspective and do not use either of the terms were not included in this review. As a consequence, often cited papers such as the articles from Bauman et al. , or Bryan et al.  have not been identified in our search. Bauman et al. , for example, examined correlates and determinants of PA with the goal to develop a multi-level ecological model for understanding the causes of PA behavior. The authors aimed to develop a model that also takes into account how etiological factors differ between PA domains, areas of life in which activity is observable, and country, age, sex, ethnic origin, and socioeconomic status. Bauman et al.  concluded that individual-level factors such as age, sex, health status, self-efficacy, and previous PA are consistent correlates of PA, but the physical and social environments, such as policy, economic conditions, societal norms, urbanization, industrialization, or interpersonal relations are also important determinants of PA behavior. Not least, they also argued that a genetic and evolutionary physiological component has to be considered when analyzing the correlates of PA behavior. Against this background, the model of Bauman et al.  can clearly be considered a biopsychosocial model despite the fact that they did not mention this term in their paper.
Similarly, the transdisciplinary model of PA behavior by Bryan et al.  also offers an exemplary approach for a multi-perspective analysis of PA. Bryan et al.  modeled the complex interactions of different components for behavioral change using the example of an intervention that promotes engagement in PA. In this regard, they explicitly considered the individuality of people and thereby individual differences as a reason to employ a transdisciplinary approach. They not only included a physiological, a genetic, a psychological, and a behavioral component into their theoretical model but also considered bidirectional relationships between components in order to explain (possible) effects and determinants of PA. However, this is a fundamental difference to the model of Bauman et al. , they completely abandoned the social component in their transdisciplinary approach. Although the article is generally very informative and helpful for designing intervention programs, the lack of a social component is a relevant shortcoming, as PA behavior is highly influenced by social and environmental factors.