Lili Kokai

The current thesis set out to design, implement, and evaluate a theory- and evidence-based mHealth intervention to decrease risk for a leading cause of mortality and disability worldwide: cardiovascular diseases (CVDs). Specifically, this thesis aimed to conduct an innovative intervention to reduce risk of CVDs through behavioral risk factors, such as insufficient physical activity, accompanying cardio metabolic risk factors, for example overweight, and psychosocial risk factors, such as stress. Growing evidence suggests that interventions based on high-quality behavioral theory (i.e. that incorporates up-to-date insights from the vast knowledge on behavior change that is already available and has been found successful in predicting behavior at least observationally in previous studies) may lead to larger effects in health contexts than non-theory-based interventions. Interventions that are based on such a health behavior theory and then systematically link theoretical constructs to evidence-based behavior change techniques are hypothesized to be most effective of all. Therefore, this thesis set out to design a theory- and evidence based intervention to reduce CVD risk. Individual-level health promotion interventions that are based on theory are typically based on a prominent social cognition theory which describes behavior as the result of deliberative psychological processes. While interventions based on such theories have generally been shown to be effective in changing behavioral intentions, they often stop short of changing actual behavior. The insights from dual process theories might help to reduce this ‘intention-behavior gap’, as they account for both automatic processes and deliberative processes.

In preparation of the selection of a dual process theory to underlie our intervention, we have decided to compare the construct of self-control both theoretically as well as empirically in psychology and economics, presented in chapters 2 and 3. Chapter 2 presented a narrative review of the theory and measurement of self-control in psychology and economics to develop a common conceptual framework. Based on the reviewed literature, we were able to show that self-control can be conceptualized along three main characteristics: stability (trait versus state), process (impulsivity versus inhibition), and enactment (avoidance versus resistance). This framework highlights the multidimensional nature of self-control and will aid intervention researchers to select theories and measurements of self-control that are most appropriate for their health outcome of interest. In chapter 3, a cross-sectional study on the relationship between several measures of trait self-control, and their relationship with modifiable risk factors for CVDs was performed. We used several measures of trait self-control that are generally considered to capture inhibitory processes and several measures that capture impulsive processes. In general, we found both higher inhibition and impulsivity to relate to more healthy behavior, less unhealthy behavior, and some healthier cardio metabolic outcomes, with inhibition showing these patterns for more outcomes than impulsivity. However, not all findings, especially those concerning cardio metabolic outcomes, showed consistent patterns, and the variance accounted for by trait self-control measures was small for all outcomes examined. Our results indicate that both inhibition and impulsivity influence health independently and simultaneously.

Chapters 4 and 5 adopted a qualitative approach to assess the needs and preferences and the perceived determinants of physical activity of the intended population of our subsequent intervention. In chapter 4 women who have experienced a hypertensive pregnancy disorder were asked about their needs: the extent to which they struggle to participate in cardiovascular health promoting behaviors, the extent to which they plan to make positive changes to these behaviors, and the extent to which they are interested in participating in an app-delivered program targeting these behaviors. Second, these women’s preferences regarding the delivery of app-based cardiovascular health promotion, i.e., their wishes regarding app content, functionalities, and interface, were examined. Women’s primary need for health behavior promotion pertained to their fat and sugar intake and physical activity. Their next priority was to gain better means to manage their mental health. That the primary needs of women are closely linked to CVD risk emphasize the need for interventions that target these behaviors in this priority population. As a healthy lifestyle, such as engaging in physical activity, has been linked to improved mental health, future interventions could target multiple needs simultaneously. Most women preferred the app-based intervention to include, in descending order: the tracking of health-related metrics, an interactive platform, the use of behavior change strategies, the provision of information, and personalization.

In chapter 5, the perceived determinants of these women’s physical activity were qualitatively assessed, and the themes that emerged were used to examine the relevance of a dual process theoretical framework. Participants perceived a wide range of facilitating and hindering factors to impact their physical activity. Thirteen themes emerged from the qualitative analysis, which were matched to four overarching themes: motivational processes (future health, perceived ability, attitude, future reward or regret, physical appearance, doing it for others), volitional processes (scheduling, planning), automatic processes (affect, stress), and environmental factors (time constraint, social support, physical environment). These themes had reasonable correspondence with the overarching motivational, volitional, and automatic processes described in the integrated behavior change (IBC) model. In addition, our results indicate that this model could be extended with the motivational process of future reward, or regret and environmental factors.

The study design of the resulting intervention is presented in chapter 6, while its short-term effects and process evaluation are presented in chapter 7. The efficacy of the eight-week intervention was tested using a three-condition randomized controlled trial (RCT) delivered through a purpose-built app, the i2be app, in women with a prior hypertensive pregnancy disorder. The intervention was based on the IBC model, which outlines the motivational, volitional, and automatic processes that lead to physical activity. Following stratification on baseline factors, participants were randomly allocated to one of three conditions – the information condition, which was meant to mimic usual care, the motivation condition, which targeted motivational processes, and the action condition, which targeted all three of the processes described by the IBC model. The primary outcome was weekly minutes of moderate-to-vigorous physical activity, as measured by an activity tracker (Fitbit Inspire 2). Secondary outcomes included the weekly average of Fitbit-measured daily resting heart rate, and self-reported BMI, waist-hip ratio, cardiorespiratory fitness, and subjective well-being. Tertiary outcomes included self-reported variables representing motivational, volitional, and automatic processes. Outcome measures were assessed at baseline, immediately post-intervention, and will be assessed at 3 and 12 months post-intervention. A process evaluation was performed based on program fidelity and acceptability measures immediately after the intervention. Efficacy was determined by available case analysis, and the mechanisms by which the behavior change techniques were hypothesized to lead to physical activity were tested. The action condition was unsuccessful in increasing physical activity relative to the information condition (usual care) or the motivation condition. We found some tentative evidence that the action condition worked better for those with low physical activity at baseline – arguably the group that has most to gain from such interventions.

There are several possible reasons for the lack of effect we find for the full sample. Physical activity levels in all conditions, including information (i.e., the control condition), were unusually high: they were highest at baseline (approximately 4 hours per week), with over two thirds of participants exceeding 2.5 hours per week. Fitbit-measured physical activity at baseline was higher than self-reported physical activity over the last month. These findings suggests that, at least at baseline and perhaps throughout the intervention, the physical activity levels of participants might have been higher than prior to enrolling in the study. We identified health promoting changes in outcomes across all conditions, most notably BMI, consistent with this hypothesis. A possible cause of such potential change in physical activity could have been the features of the information condition, which were also available to the other two conditions, therefore possibly not having been an appropriate usual care benchmark. In particular, we might have underestimated the effect of the Fitbit device and app and the basic version of the i2be app on physical activity, which might have crowded out effects that may have otherwise arisen from the behavior change techniques included in the motivation and action conditions. Further, the lack of success of the action condition to significantly influence automatic processes may also have played an important role in the overall lack of effect, since the IBC model hypothesizes these processes to be influential in the intention-behavior gap regarding physical activity.