Many recent studies have shown the crucial role that motivated behaviors, such as regular exercise, following a healthy diet, and not smoking, can play in the maintenance of health 11; many people at health risk have the means readily at hand to improve their condition, assuming that they are willing to act. In fact, the failure of many people to adhere to healthy behaviors represents a public health problem, one with many causes but for which a considerable part is motivational in nature. This highlights the need for a clearer understanding of what motivation is and how to facilitate it in the context of health behavior. Motivation refers to the psychological forces or energies that impel a person toward a specific goal. However, for many years, motivation was viewed in a one-dimensional way, as varying only in amount or quantity. Introducing the issue of quality of the motivational drive, Deci and Ryan 1213 developed the Self-Determination Theory (SDT) which distinguishes between amotivation (lacking any intention to engage in a behavior), extrinsic motivation (where the behavior is engaged in order to achieve outcomes that are separable from the behavior itself) and intrinsic motivation (where the behavior is engaged in for the enjoyment and satisfaction inherent in taking part). In addition, SDT distinguishes between qualitatively different forms of extrinsic motivation, by contrasting autonomous or self-determined vs. controlled or non-self-determined types of behavioral regulation. This distinction is represented as a continuum and is characterized in terms of the degree to which the regulation of a behavior has been internalized so that it is engaged in with a true sense of volition and choice. Motivation is autonomous to the extent that a person's perceived locus of causality is internal (i.e. the perceived source of initiation and regulation for motivated behaviors emanates from the self). Motivation is controlled to the extent that people act because they feel pressured or compelled to do so, either by others or by themselves, and this involves having an external perceived locus of causality 14.

According to Deci and Ryan, intrinsic motivation is linked to greater productivity, creativity, spontaneity, cognitive flexibility, and perseverance 12. However, most human behaviors are not intrinsically motivated, which highlights the importance of studying extrinsic forms of motivation. Four types of extrinsic motivation have been described, which can be located on a self-determination continuum 15. External regulation is at the lower end of the continuum. It means doing something in order to gain a reward or to avoid a punishment administered by others. Introjected regulation involves an internal feeling of obligation, a need to act in order to avoid feelings of guilt (although the pressure stems from inside, the individual does not feel free regarding the behavior). Identified Regulation concerns doing something based on the value of its consequences. In this case, the individual feels free to act and does so because the outcomes are personally important. Finally, when a behavior is coherent with the person's other values, personality schemas and sense of self, the regulation is fully self-determined and said to be integrated. SDT proposes that human motivation, vitality, development, and psychological adaptation can be explained by the process of internalization, that is, movement along this continuum 13. A model describing internalization and human motivation is presented in Figure 1.

Autonomous behavior is an expression of one's self and is undertaken with a full sense of choice, accompanied by an internal perceived locus of causality and a sense of true volition. Importantly, according to this theory, autonomously motivated behaviors are more likely to result in meaningful, long-lasting changes. Conversely, behaviors that are controlled by rewards and punishments or by self-imposed pressures are predicted to last only as long as the contingencies or pressures remain in place.

Fortunately, according to SDT, formerly controlled motivation can be internalized and transformed into autonomous motivation, if supportive conditions are in place. In fact, this contemporary theory of human motivation focuses on the psychological nutriments that engender adaptive motivational, behavioral, cognitive, and affective outcomes, by specifying contextual variables that facilitate (or hinder) these processes. The way a person acts in a particular setting cannot be attributed only to individual differences; contextual variables also exert a significant influence. Thus, SDT also considers the influence of the social environment on motivation. According to Deci and Ryan 13, although people have an inherent propensity toward maintaining their well-being, this natural tendency can be thwarted by conditions that frustrate the satisfaction of three basic psychological needs. These needs are for autonomy (feeling volitional and feeling choice and responsibility for one's behavior), competence (feeling that one can accomplish the behaviors and reach the goal) and relatedness to others (feeling understood, cared for and valued by significant others). In the context of health, socio-environmental conditions that facilitate the satisfaction of these needs will promote the internalization of protective and preventive health behaviors so that they are engaged in autonomously and more likely to be maintained in the long term. A recent study in the context of exercise adherence clearly indicated that fulfillment of these three basic needs was related to more self-determined motivational regulations 16. It can generally be stated that people need to feel a sense of choice and volition with respect to their health-related goals; they need to understand how to attain these goals and feel that they can be effective in carrying out the necessary actions; and feel respected and cared for by practitioners and important others. This means that people not only need to feel that they can carry out a certain behavior (confidence, efficacy, competence), they also need to feel that they are fully responsible for initiating and maintaining that behavior and that they are doing so willingly (autonomy, self-determination, responsibility).

According to SDT, a person will develop and maintain more self-determined motivation when the personal context around them is autonomy supportive. The idea of autonomy support refers to eliciting and acknowledging peoples' perspectives, supporting their initiatives, offering choice/options, and providing relevant information, while minimizing pressure and control. Previous research has demonstrated that when physicians are perceived by their patients as being autonomy supportive, patients report greater self-motivation for treatment adherence 17. Another recent study, designed to test a SDT intervention for motivating tobacco cessation in a clinical trial 18, showed that intervention participants perceived greater autonomy support and reported greater autonomous and competence motivations than did controls, supporting the causal role of autonomy support in the internalization of more internal forms of regulation.

According to SDT's basic tenets, successful maintenance of weight reduction would occur when people chose eating and exercise behaviors because they personally value weight loss maintenance and its health benefits. Lasting behavior change necessary for maintenance depends not on complying with demands for change but rather on accepting the regulation for change as one's own. In other words, it requires internalizing values and regulation of relevant behaviors and then integrating them with one's sense of self so they become the basis for autonomous regulation. Accordingly, successful weight loss and long-term maintenance would not be achieved if reasons for it were mostly controlling (e.g., because the doctor insisted or based on a strong desire to be thin, according to social norms). Being autonomous in one's relevant actions, that is, having an internal perceived locus of causality, is the crucial predictor of maintained behavior change 19. In a study specifically designed to study the effects of motivational factors during a weight control program, results showed that the degree of autonomous motivation predicted not only attendance at weekly meetings and weight loss during the program, but, more importantly, maintenance of weight loss at the 23-month follow up 20.

The role of exercise in weight management has also been analyzed in relation to SDT. In a recent study on psychosocial predictors of weight management, increase in intrinsic motivation for physical activity was the strongest predictor of long-term weight change, even after adjusting for initial weight-loss 21. The authors stressed the importance of ensuring that individuals take on physical activities that they intrinsically enjoy, feel competent at, and that contribute positively to their sense of autonomy as key factors for greater success in the difficult task of long-lasting weight control.

A common theme emerging from the problem of exercise adherence concerns the role of individual's reasons for exercising (participation motives) in determining long-term adherence to regular physical activity. Markland's study findings 22 support the idea that extrinsic motives such as exercising for weight control or because of a doctor's exercise prescription may be perceived as controlling thereby undermining self-determination and leading to a lack of enjoyment of exercise. On the contrary, intrinsic motives, such as exercising for enjoyment may be perceived as informational, enhancing perceptions of self-determination, suggesting that controlling exercise regulations may lead to a greater number of relapses from exercise compared with more self-determined types of exercise regulation. Importantly, exercisers in the maintenance stage of change display significantly more self-determined motivation to exercise than those in the preparation and action stages 23. Also, others have shown that adherence to exercise in individuals participating in fitness classes is higher when intrinsic motives related to enjoyment and feelings of competence were reported, compared to when body related outcomes (conceptualized as extrinsic) are the primary motivation 24. Another study on perceived autonomy support for exercise concluded that participants who exercised regularly report modest to large changes in relatedness and competence need satisfaction and intrinsic motivation over time, providing theoretically-sound insights into reasons why people persist with exercise behaviors 25.

The mechanisms that promote self-determined motivation for exercise are now under investigation. Overall enjoyment, perceptions of competence, and intrinsic reasons for participation appear to play a central role in the maintenance of physical activity behaviors 23. In a large scale RCT, increases in enjoyment mediated physical activity in female adolescents involved in a school-based intervention 26. In fact, while longitudinal studies are required to better understand the processes underlying maintained exercise behavior, previous results support Markland and Hardy's 27 proposition that an individual's motivational focus needs to shift from extrinsic to intrinsic for long lasting results. New models are needed to understand this shift. We argue that understanding the motivational basis of long-term weight management and exercise requires adopting a more general model of motivation than that offered by the behaviorist tradition, which focuses on controlling a person's behavior with rewards and punishments and on the external determinants of behaviors. It also requires adopting a model more general than that offered by the social cognitive tradition, characterized by a higher emphasis on expected outcomes and enhancing patients' sense of confidence or self-efficacy (quantity of motivation) but not focusing on the quality of motivation, sense of autonomy, and locus of causality 28. Self-Determination Theory provides such a model.

Figure 2 presents the self-determination model of health behavior change and the expected relationships among its key constructs. Central to the model is autonomous self-regulation for behavior change. Autonomy support, both experimentally manipulated and as perceived by patients, and also a participant's general autonomy orientation are predicted to enhance autonomous self-regulation and perceived competence. Autonomous self-regulation and perceived competence are in turn expected to increase maintained change of the health-risk behaviors 29.

A recent study 18, aimed at prolonged abstinence from tobacco, was the first to demonstrate that an intervention based on the SDT model facilitates the internalization of motivation, resulting in expected behavioral outcomes. This extends in many ways the previous empirical support for the model found in predicting maintenance of glucose control for diabetes 17, and maintained tobacco abstinence for smokers treated by primary care physicians 30, showing that only when perceived competence is accompanied by the experience of autonomy will it motivate sustained change.

The SDT health care model was formulated to account for the promotion of all health-relevant behaviors, including nutrition and physical activity. Edmunds and colleagues demonstrated that the basic tenets advanced by SDT are amenable to manipulation and revealed the potential utility of SDT to shape behavioral interventions targeting the promotion of healthier and more physically active lifestyles 16. Given the difficult task of long-lasting weight control, an intervention that could significantly increase the percentage of people able to successfully manage their weight in the long-term, through voluntary, stable changes in diet and physical activity, would represent an important contribution.

We selected SDT as the theoretical basis for this study because it is the only empirically-derived theory of motivation which posits that perceived autonomy (or self-determination) is essential for maintained behavior change and because there are validated psychometric instruments for each construct of the theory. Thus we can test an intervention as well as testing a theory-based process model, assuming the need to treat a theory as a dynamic entity whose form and value rests upon being rigorously applied, tested, and refined 31.

Randomized controlled trial, analyzing 3-year change in primary outcomes, consisting of a 1-year behavior change intervention and a 2-year follow-up period with no intervention. Participants (n = 259) entered the study in 3 annual cohorts and each cohort was split into 2 randomly-assigned groups: one intervention group and one comparison group. The intervention group attended about 30 group sessions for approximately one year. The comparison group received a general health education curriculum based on several 3 to 6-week long educational topics (e.g. preventive nutrition, stress management, self-care and effective communication skills, among others). The Technical University of Lisbon – Faculty of Human Kinetic Ethics Review Board approved the study.

In this trial, assessments were planned to occur at baseline, 4, 12, 16, 24, and 36 months. Besides physical activity, diet, and physiological and psychosocial variables (all described below), demographic variables including menopausal status, age, education level, job or occupation, and socio-economic status were assessed via self-report.

Weight was measured twice, to the nearest 0.1 kg (average was used) using an electronic scale (SECA, model 770, Hamburg, Germany) and height is also measured twice, to the nearest 0.1 cm (average was used). Body mass index (BMI) in kilograms per squared meter was calculated from weight (kg) and height (m). Body composition was assessed by dual energy x-ray absorptiometry, DXA (QDR-1500; Hologic, Waltham, MA, USA, software version 5.67). Assessment of anthropometric measurements also included body circumferences, total body, trunk, and abdominal fat mass, total body and appendicular fat-free mass. Resting metabolic rate was measured using the MedGem portable device (MedGem^®, Health TechTM Inc., Golden, CO, USA).

Physical activity and exercise was measured objectively with the GT1 accelerometer (Actigraph LLC, Pensacola, FL, USA), which quantifies activity counts for every day of the week and estimates free-living PA and energy expenditure at different intensity levels. Participants also concurrently carried a Yamax Digi-Walker SW-200 pedometer (New Lifetyles, Lee's Summit, MO, USA) for assessment of daily steps. This pedometer is a highly accurate and a reliable step counter and is effective for both walking and running 4041. In addition, the 7-day Physical Activity Recall (7-d PAR) was used, in interview format, asking subjects to list all activities of moderate, hard, and very hard intensity, as well as daily walking minutes 4243. For this calculation, the time spent in each category (hours) is multiplied by body weight and a caloric, or MET value (based on intensity) to yield total exercise energy expenditure.

Dietary intake was assessed with three 24-h diet records (randomly assigned days each testing period, including one weekend and two week days). Diet records are analyzed for energy and nutrient intake using Food Processor SQL (Nutrition Analysis Software Version 7.4, ESHA Research, Salem, OR, USA).

Participants completed a comprehensive battery of psychosocial measures covering several areas considered as relevant to weight management 44. This was conducted in standardized conditions of comfort and silence, with a study technician attending every assessment period. Psychosocial areas included previous diets and weight history variables taken from a diet/weight history questionnaire developed specifically for this study. Weight outcome evaluations were assessed by 4 questions derived from the Goals and Relative Weights Questionnaire (GRWQ) 45, body size dissatisfaction was assessed by the difference between self and ideal body figures selected from a list of 9 female silhouettes of increasing size 46. Body image was further evaluated with the Body Shape Questionnaire (BSQ) 47 and the Physical Self-Perception Profile (PSPP) 48. Physical activity determinants were assessed with the Exercise Perceived Barriers (EPB)49, and Exercise Self-efficacy (ESE) 50 scales. Dietary restraint, disinhibition and hunger were assessed with the Three-Factor Eating Inventory 51. Psychological states and traits were also considered. Depression status was assessed with the Beck Depression Inventory II (BDI-II) 52 and Anxiety was evaluated with the State-Trait Anxiety Inventory (STAI) 53. Self-esteem was assessed with the Self-Esteem/Self-Concept Questionnaire 54. Quality of Life related to weight was measured with the Impact of Weight on Quality of Life Questionnaire (IWQOL) developed by Kolotkin et al. 55. Health-related quality of life was assessed with the Medical Outcome Survey Short-Form (SF-36) developed by Ware 56. Social support for exercise was measured with the Exercise Social Support Scale (ESS) 57. The following psychosocial questionnaires related to SDT were also included, in order to test the intervention model (see Figure 1).

General causality orientations were assessed with the General Causality Orientations Scale (GCOS) 58. Exercise intrinsic motivation was measured with the Intrinsic Motivation Inventory (IMI) 27, adapted to specifically measure the dimensions of exercise interest/enjoyment, perceived competence, effort/importance and pressure/tension.

The Locus of Causality for Exercise Scale (LCE) 59 assessed locus of causality for exercise. The Health Care Climate Questionnaire (HCCQ) 20 measured participants' perceptions of the degree of autonomy support (vs. controllingness) of the relevant health-care providers. The Exercise Self-Regulation Questionnaire (SRQ-E) 60 assessed domain-specific individual differences in types of motivation or regulation (external, introjected, identified, and integrated). Finally, the Exercise Motives Inventory (EMI-2) 6162 assessed exercise participation motives on five dimensions (psychological, interpersonal, health-related, body-related and fitness-related motives), including 14 subscales.

At baseline and treatment's end (12 months) individual interviews of approximately 45 minutes were also conducted. The addition of qualitative information to quantitative data (i.e., psychometric instruments) is an important and innovative aspect of this study's assessment plan. Due to space limitations, interviews are not described in detail herein.

Analysis will be conducted for completers-only and using an intent-to-treat model. Baseline data imputation for missing scores at follow-up will be employed by default. Despite limitations of any imputation strategy, this method is conservative and has been recommended before 63. The use of these models is very important in clinical trials where completion is less than perfect, since completers-only analyses can be severely biased and underpowered 44.

For analysis of moderators and mediators, we will follow recommendations by Kraemer et al. 10. Generally, for each primary outcome, linear models with be conducted to test for 1) main effect of intervention (I), 2) main effects of each moderator or mediator (M), 3) interactive effects of M and I, 4) overall effect of intervention (plus high vs. low M) and 5) effect of intervention for sub-groups of subjects at specific levels of M. A moderator will be defined as a baseline variable (thus uncorrelated with intervention) that interacts with intervention assignment to produce significant effects. The overall effect of intervention ("additive" effect of intervention and M) will represent the actual (clinical) impact of intervention participation in the presence of M. Finally, profiles of subjects with particular values of M may be identified who are responsive to this particular SDT-based intervention.

Measures of central tendency and distribution will be examined for body weight and psychosocial variables at the different measurement periods. Repeated measures ANOVA and ANCOVA General Linear Model will be used to assess change in dependent variables and to test for within-and between-group differences. Multiple regression analyses and recursive partitioning (signal detection methodology) 64 will be employed to identify characteristics/profiles associated with success, following the principles enunciated above.

Sample size calculations were estimated for primary outcomes for a power of 0.8, and a two-tailed p < 0.05. At 1 year, for physical activity, we assume an increase of 300 ± 500 METs-min/sem (intervention, mean ± SD) and 50 ± 400 METs-min/sem for comparison's. A difference effect size (ES) of 0.56 is obtained, for which a sample of 52 participants/group is needed. For changes in body weight (% change of initial) and relative body fat (% change from initial) estimated ES are large (>0.83) and would require considerably less participants/group. Difference ES for some psychosocial variables were also calculated, with estimates (means and SD) based on our previous work 65, and range from 0.67 to 0.83. At maximum, 36 participants/group would be needed to detect significant differences in these variables between intervention and controls. Assuming a dropout rate of 20%, 120 participants/group will remain at 1-year (our primary endpoint), enough to detect even small differences in PA. The lowest ES observed (for PA, ~0.5) would require ~60 participants/group. In the behavioral sciences, "moderate" and "strong" ES for associations between variables (R²x100 or percent variance accounted) have been set at 5% and 10% respectively, which is equivalent to correlations of 0.22 and 0.31 66. Thus, our sample size will be adequate even for the lowest (i.e., "moderate") association levels.