Randomisation and procedure
The study can be divided into five phases: (1) the phase from recruitment to randomisation, (2) EMA-only week 1, in which EMA data on eating behaviour and cognitions will be obtained, (3) the Intervention Phase; in which participants will actively follow a diet, either accompanied by the Think Slim intervention (immediate intervention group) or not (diet-only control group), (4) EMA-only week 2: a second week in which EMA data on eating behaviour and cognitions will be obtained, and (5) the Follow-up (FU) Phase: the period covered by assessment at 3 and 12 months post intervention, and by the reception of the Think Slim intervention by the diet-only control group. In each phase, different modules of the Think Slim app will be active. The flow of the trial, measures and intervention components are graphically depicted in Fig.
1. Each phase will be described in detail below. During the trial, participants are allowed to contact the research staff via e-mail and telephone to discuss any technical or personal issues that might be of influence on study participation. In addition, telephone contact is scheduled prior to the onset of each new phase. Care will be taken to avoid biasing participants during this communication, by referring to the online training and app manual whenever content-related questions arise.
Phase 1: from recruitment to randomisation
After initial response to the advertisements, interested candidates will be requested to complete an online screening questionnaire assessing inclusion and exclusion criteria. Eligible participants will receive further study information via e-mail and telephone. This will allow candidates to ask questions and to consider their study participation well before they agree to participate. Participants who are not eligible for participation will be informed via e-mail about the reason(s) for exclusion and will be provided with an opportunity to correspond with the research staff.
After agreeing to participate, participants will provide digital informed consent. Subsequently, participants will be contacted to plan a 1-day technical training via the app shortly afterwards, and will be sent a manual for the app via postal mail and via e-mail. The purpose of the technical training is to ensure that the app is working and to minimise the influence of differences in technical aptitude.
App functionality during the technical training
During the technical training, the EMA module and Graphical Overview Feedback module of the app will be active. The EMA module of the Think Slim app will be used to obtain real-time data of eating-related variables. Items that are part of the EMA questionnaires include measures of food desire strength, specific food desires, emotions, cognitions, locations, activities, social company and food consumption. During the training, EMA will be performed at two separate instances: (1) eating events (event sampling), and (2) eight times pseudorandomly throughout the day (time-contingent sampling). Sampling of an eating event will be initiated by the participants themselves, whereas time-contingent sampling points will be generated by the app. Note that cognitions and food consumption will only be assessed during eating event sampling. Otherwise the eating event samples and time-contingent samples are the same. The first time-contingent sample of the day will also assess sleep duration and quality. Beverage consumption will be assessed in a separate assessment, which will occur prior to bedtime.
In addition to the EMA module, graphical overview feedback will be accessible in a separate section of the app. The feedback will represent pie charts of the activities and locations assessed at eating event samples by the participant, separately for foods that are considered healthy and unhealthy. In addition, a line graph of desire strength and strength of various emotions, presented relative to time, will be available. The graphical overviews will continuously be updated, as soon as new EMA data is entered into the app. To generate the overviews, at least one eating event sample is necessary.
Randomisation
After completing the technical training, all participants will be contacted by the researchers to discuss possible issues that occurred during the training. Subsequently, randomisation will take place. Participants will be randomly assigned to either the immediate intervention group or the diet-only control group using computer-generated randomisation via Research Randomizer (
www.randomizer.org). More specifically, randomisation is pre-stratified by age group (18–30; 31–40; 41–50; 51–60) and gender. For each age group, separately for male and female participants, each participant will be assigned a randomly generated number. Half of the participants with the largest randomly generated numbers will be assigned to the immediate intervention group, whereas the other half will be assigned to the diet-only control group. Note that participants who indicate as being unavailable during a specific time period, due to for example holidays, will be manually allocated to the condition that matches with their availability. Blinding participants and researchers is not possible in the current setup.
Phase 2: EMA-only week 1 (week 1)
The trial starts with an EMA-only week. At the start of this week, all participants (immediate intervention group and diet-only control group) will fill out a baseline assessment consisting of several self-report questionnaires (specified below, under ‘Measures’) about eating styles, eating disorder pathology, general psychological distress, self-esteem and dysfunctional thinking. In addition, participants will obtain an independent measure of their BMI. No further intervention will take place.
App functionality during the first EMA-only week
During the entire week, the EMA module and Graphical Overview Feedback module of the app will be active, identical to the technical training.
Phase 3: Intervention Phase (weeks 2–7)
After the first EMA-only week, the Intervention Phase starts. This phase has a total duration of 6 weeks (weeks 2–7).
Diet
During the Intervention Phase, all participants (immediate intervention group and diet-only control group) are asked to actively follow a calorie-restrictive diet of their own choice. Supervision of the diet by a dietician or healthcare professional is not allowed. Instead, all participants select a diet coach (friend or family member). Participants will also develop an exercise plan. General information about dieting and exercising is provided as part of two online sessions in the first week of the Intervention Phase. At the end of the first week, participants are asked to finalise their choice. Aside of the two online sessions, no further information about selecting a diet is provided.
Cognitive intervention (for immediate intervention group only)
The cognitive intervention will be aimed at changing obesity-related dysfunctional thinking and at increasing self-esteem. It consists of ten individual online sessions on a computer (cCBT), and a module in the app. The computer sessions are included because CBT requires explanation of the theoretical model and therapeutic concepts that are too comprehensive to provide via an iPhone app due to screen size limitations that prevent a convincing digital presentation of the intervention. Therefore, the online sessions provide participants with the necessary information on the cognitive model and cognitive techniques, whereas the app is suitable to practise these skills in daily life.
There are a total of ten online sessions of approximately 25 minutes that will be delivered via an in-house web-environment for online therapy and questionnaire research (SOTO). To start each session, participants receive an e-mail with a link to the SOTO web-environment. The sessions are based on several CBT for weight loss protocols [
45,
58,
59]. They are designed to be provided in addition to a standard calorie-restrictive diet. Table
1 contains a brief summary of each session’s contents. In the first week of the Intervention Phase, five sessions will take place. In these sessions, participants will determine reasons for losing weight, set dietary goals, choose a diet, make an exercise plan, and make a list of techniques to distract from food craving. All information except the choice of diet and exercise plan will be accessible for review in the app. Furthermore, participants will be familiarised with the cognitive model and basic cognitive techniques. They will also learn how to identify and evaluate dysfunctional cognitions, and how to replace these with alternative cognitions that are more helpful towards eating behaviour and dieting. For the first week, all five sessions will need to be completed at the end of the week. Participants who do not complete all five sessions will be considered dropouts. Participants will be sent reminder e-mails in case they lag behind and will be contacted by the research staff in case they are close to being considered dropouts.
Table 1
Summary of the contents of the computerised cognitive behavioural therapy (cCBT) sessions
1. | Introduction | Information on weight-gain and weight-loss Introduction of the cognitive model and the concept of dysfunctional thinking Information on how to select a diet, diet coach and exercise plana |
2. | Motivation | Personal reasons for losing weightb Realistic weight loss goalsb Identifying and responding to sabotaging cognitions about dieting Personal list of sabotaging cognitions about dietingb |
3. | Coping with food desire | Information on the difference between hunger and food desire Tips on how to deal with food desire Personal list of distraction techniquesb |
4. | Identifying dysfunctional cognitions | Identifying dysfunctional cognitions about eating Introducing the concept of dietary temptations Instruction on how to do this in the app |
5. | Responding to dysfunctional cognitions | Introduction of socratic questioning Information on how to construct functional alternative cognitions Instruction on how to do this in the app Choosing a diet and exercise plana,c |
6. | Thinking errors | Summary of basic CBT skills Introduction of concept of thinking errors Instruction on how to do this in the app |
7. | Interim evaluation | Evaluation on progress so farc Increasing motivation Learning to compliment oneself |
8. | Self-esteem: body | Positive body exposure |
9. | Self-esteem: person | Designing a positive portrait of the selfc |
10. | Evaluation and relapse prevention | Summary of interventionc Evaluation of intervention Creating a relapse prevention planc |
From the second week until the last (week 7) of the Intervention Phase, participants will complete additional cCBT sessions at the end of each week (sessions 6–10). In these sessions, cognitive skills are reviewed, and participants are encouraged to stay motivated. Furthermore, self-esteem is addressed in two sessions, focusing on the body and on personality characteristics. In the final session special attention is paid to evaluation and relapse prevention. All online cCBT sessions will be completed without therapist involvement. Participants are allowed to lag behind one session, and will be considered dropouts if more than one session is not completed on time. At the beginning of each session, dysfunctional thinking will be assessed via a self-report questionnaire. Participants in the diet-only control group will not receive any cognitive intervention during the Intervention Phase.
App functionality during the Intervention Phase
Participants in the diet-only control group will not be able to use the app during the Intervention Phase. For these participants, all app functionality will be blocked. Instead, the app will display the number of days remaining until the end of the waiting period. Eight days before the end of the waiting period, the app will send a notification informing participants that the end of the waiting period is near.
For participants in the immediate intervention group, app functionality in the first week of the Intervention Phase will consist of the EMA module and the Graphical Overview Feedback module, similar to the training and the first EMA-only week. In addition, the Adaptive Feedback module will be active. The Adaptive Feedback module provides participants with feedback if they are at risk for overeating. The feedback message consists of a generic warning and a behavioural advice. Detection of a participant’s ‘risky’ moments is based on the answers provided by the participant on the EMA items. To estimate when the participant is likely to eat something unhealthy in the time-period directly following a completed time-contingent sample, the sample data is compared to a set of allocated pre-existing rules that indicate what combinations of states of variables (e.g. scoring high on desire + being at home + feeling bored + feeling calm) are predictive of unhealthy eating. If there is a match between the time-contingent sample and one of the rules, the participant receives adaptive feedback messages via the Think Slim app.
To provide adaptive feedback, the Think Slim app uses 65 rules that were obtained via a computational decision-tree algorithm (adapted for EMA data) used on a dataset from a previous EMA study that had an almost identical EMA protocol (Boh B, Jansen A, Clijsters I, Nederkoorn C, Lemmens L, Spanakis G, & Roefs A. Indulgent thinking. Ecological Momentary Assessment of Overweight and Healthy-weight Participants’ Cognitions and Emotions, submitted). Each of the 65 rules represents states of eating-related variables that were most frequently reported together right before a healthy or unhealthy eating event, as reported by participants of that study. More specifically, participants were divided to six groups according to their rule-triggering occurrence and each group is represented by a set of rule that describes 80 % of the eating behaviour of participants in the group (thus removing rules with low occurrence and keeping only those with high predictive value). For the current study, each participant will be allocated one of the six rule-triggering groups based on EMA data from the first EMA-only week. At the end of the third week of the Think Slim intervention, EMA sample data from the first 3 weeks of the Intervention Phase will be used to update the allocation of participants to the rule-triggering groups.
From the second week of the Intervention Phase until the end of the intervention, the app functionality will be further expanded. First, participants will be instructed to also assess moments of dietary temptation (event sampling) in the app. Participants will be instructed in an online cCBT session about what qualifies as a dietary temptation. The EMA items assessed during moments of dietary temptations are similar to those assessed during an eating event, except for the food consumption item: food desire strength, specific food desires, emotions, location, company, and activities. Second, the Cognition Evaluation module will be active. The purpose of this module is to identify and evaluate dysfunctional cognitions in daily life. Table
2 contains an overview of the cognitions included in the app and whether or not these are considered dysfunctional. The module designed to work in conjunction with the online cCBT sessions: the module requires knowledge of the basic cognitive skills that are provided in online sessions 1–5.
Table 2
Cognition types included in the Think Slim app and the categories they belong to
Neutral | Description of an eating event | ‘Time for breakfast!’ |
| Hunger | ‘I am very hungry, time to eat.’ |
| Desire and taste | ‘I really want this chocolate bar.’ |
| Energy needed | ‘I won’t have time to eat later.’ |
Functional | Healthy intention | ‘Apples are good for me.’ |
| Successful control | ‘Everyone’s eating cake, but I’m going for a pear.’ |
Dysfunctional | Negative emotions | ‘I feel awful, maybe this helps.’ |
| Positive emotions | ‘I’m trying to relax with snacks.’ |
| Social activities and pressure | ‘Everyone’s eating cake. I will join them.’ |
| Reward | ‘I deserve this after all the hard work!’ |
| Control failure | ‘I just can’t resist.’ |
| Other dysfunctional cognitions | ‘Coffee should go together with a cookie.’ |
When a dysfunctional cognition is selected by a participant in the Cognition Evaluation module of the app during an eating event or dietary temptation, the participant is encouraged to critically evaluate the selected cognition via socratic questions, which aim to guide the evaluation process. For the ease of evaluation, the cognition is presented in a conditional (if-then) format. After the evaluation, the participant is asked to aggregate the information obtained in the evaluation section and use this information to formulate a helpful and realistic response to the initial dysfunctional cognition. Subsequently, the believability of the new cognition is rated. To conclude, the participant is asked to take another look at the initial cognition and rate believability again. The idea is that after evaluation, the believability – and therewith the influence – of the initial cognition decreases. All evaluated cognitions remain accessible in the app for future reference. If a functional or neutral cognition is selected in the app by a participant, the app also provides feedback. In case of an eating event, this feedback will be a generic message (e.g. ‘Enjoy your meal’.). In case of a dietary temptation, an appropriate cCBT-based feedback message will be provided to the participant. This feedback message will refer back to what was learned during the cCBT sessions if the selected cognition is considered dysfunctional.
Lastly, at the last day of week 7, a post-intervention assessment will take place during which participants will fill in self-report questionnaires identical to the baseline assessment. Participants are also instructed to obtain a second measurement of weight during week 8, under identical circumstances as during the first EMA-only week.
Phase 4: EMA-only week 2 (week 8)
After the Intervention Phase, all participants (immediate intervention group and diet-only control group) enter a second EMA-only week. This week is identical to the first EMA-only week for all participants.
Phase 5: Follow-up (FU) Phase
At the end of the second EMA-only week (week 8), participants in the immediate intervention group will be debriefed about the purpose of the trial. Furthermore, they will receive a monetary reward of €50, in coupons. After that, their FU Phase starts. The FU Phase has a total duration of 12 months. Participants are invited to fill out questionnaires (identical to the baseline and post-intervention assessments) and to obtain independent BMI measurements at two time-points during this period: at 3 and 12 months post intervention. They will receive an additional €20, in coupons, per completed FU measurement.
For the diet-only control group, the procedure after the second EMA-only week (week 8) is different. Participants allocated to this group will now receive the Think Slim intervention (weeks 9–14), followed by a third EMA-only week (week 15) and a debriefing. The content and procedure of the Think Slim intervention received by the diet-only control group is identical to that received by the immediate intervention group, except that information concerning the diet and exercise plan is not included. Furthermore, the third EMA-only week is identical to the other two EMA-only weeks. After that, the 12-month FU Phase starts for this group. The procedure of the FU Phase for the diet-only control group is identical to the procedure of the immediate intervention.
Measures
Several instruments will be used to assess the effects of the Think Slim intervention on BMI and secondary outcomes. Data are obtained by means of self-report instruments, measures taken by healthcare professionals, and EMA data. Figure
1 depicts when each measure is obtained during the flow of the study.
Body mass index
Participants are instructed to obtain multiple measurements of body weight and one measurement of height. These measurements are used to calculate BMI. To ensure objectivity, participants are asked to obtain the measurements via a healthcare professional (e.g. a general practitioner, or a physical therapist), or at the university (Think Slim research staff). Participants are further requested to obtain all measurements under identical circumstances (i.e. the same general practitioner or physical therapist).
Patterns of eating behaviour
To gain insight in (changes in) eating patterns, EMA data obtained during the first EMA-only week and second EMA-only week will be investigated.
Dysfunctional thinking
Dysfunctional thinking will be measured by believability ratings of nine frequently occurring dysfunctional cognitions about eating (e.g. ‘If everyone is eating a snack, then I should participate’.). Participants are instructed to read each statement carefully, and to answer the question ‘How much do you believe in this cognition?’ by scoring believability on a visual analogue scale (VAS) ranging from 0 to 100; anchors ‘not at all’ and ‘very much so’. Cognitions are selected based on previous research (Boh B, Jansen A, Clijsters I, Nederkoorn C, Lemmens L, Spanakis G, & Roefs A. Indulgent thinking. Ecological Momentary Assessment of Overweight and Healthy-weight Participants’ Cognitions and Emotions, submitted) and observations in clinical practice.
Eating styles
The Dutch Eating Behaviour Questionnaire (DEBQ; [
60]) will be used to measure eating styles and consists of 33 items across three subscales: external eating, emotional eating and restraint. Each item of the DEBQ is rated on a five-point Likert scale ranging from 1 (seldom) to 5 (very often). Item scores for the three subscales will first be summed separately to obtain an overall score per subscale. Then, for each subscale, the overall score will be divided by the number of subscale items to obtain the final average score per subscale. A high score on one of the subscales indicates a higher tendency to display the respective eating style. The DEBQ was previously found to be internally consistent overall,
α = 0.79 [
61]. The subscales emotional eating (
α = 0.89), external eating (
α = 0.078) and restrained eating (
α = 0.92) were also found to be internally consistent [
62].
Eating disorder pathology
The Dutch version of the Revised Eating Disorder Examination Questionnaire (EDE-Q; [
55]; Dutch translation) will be used to measure eating disorder pathology. This self-report questionnaire consists of 22 items assessing eating pathology, and 6 items assessing diagnostic behaviours such as use of laxatives and purging. The items are scored across four subscales: restraint, weight concerns, shape concerns, and eating concerns. The eating pathology items of the EDE-Q are each scored on a seven-point Likert scale. The diagnostic behaviour items are scored by participants indicating the number of times in the past 28 days that the behaviour occurred. The test-retest reliability of the English version of the EDE-Q was found to be between
r = 0.66 and
r = 0.94 for each of the subscales, whereas the internal consistency of the EDE-Q subscales was found to be good, ranging from
α = 0.70 to
α = 0.93 [
63].
General psychological symptomatology
The Brief Symptom Inventory (BSI; [
64]: Dutch translation by [
65]) will be used to assess general psychological symptomatology, and consists of 53 items. The BSI has nine subscales assessing different forms of psychological distress. Items of the BSI are scored on Likert scales ranging from 0 (not at all) to 4 (very much so). Scores for each subscale are calculated by averaging scores on the individual items belonging to that subscale. The total score for the BSI is calculated by summing all individual item scores. The BSI subscales were found to have satisfactory internal consistency, ranging from
α = 0.71 to
α = 0.88 [
66]. In addition, the test-retest reliability is high, with the lowest score reported at
r = 0.68 [
66].
Self-esteem
The Self-liking and Self-competence Scale Revised (SLSC-R; [
67]: Dutch translation by [
68]), will be used to assess self-liking and self-competence as dimensions of self-esteem, and consists of 16 items, 8 for each dimension. Items are rated on five-point Likert scales ranging from 1 (strongly disagree) to 5 (strongly agree). Overall scores per dimension are obtained by summing scores of individual items belonging to each dimension, and can range between 8 and 40. Higher scores indicate more self-competence and self-liking than lower scores. Internal consistency of the subscales of the Dutch translation of the SLSC-R was found to be good, ranging from
α = 0.78 to
α = 0.98 [
68]. Split-half reliability for the subscales was also found to be good, ranging from
r = 0.75 to
r = 0.92.
Acceptability and compliance
Acceptability of the Think Slim intervention will be measured using a questionnaire that was developed for the current study. This questionnaire includes items on user-friendliness and perceived usefulness of the Think Slim intervention. Compliance with the EMA module in the app will be measured daily during the study by calculating how many time-contingent samples were completed by participants relative to the total number of time-contingent samples that were sent by the app, and by monitoring how often the Cognition Evaluation module of the app was used.
Data analysis
Participant’s flow from screening to randomisation will be mapped out. Furthermore, pre-treatment characteristics of the sample will be explored, and baseline differences between conditions will be examined in terms of size and clinical importance. Distributions of outcome variables will be inspected via histograms and boxplots. After that, intervention- and study-compliance will be determined, followed by an evaluation of participants’ satisfaction with the Think Slim intervention.
To test the effectiveness of the Think Slim intervention, 2 (group: immediate intervention versus diet-only control group) × 2 (time: pre versus post) analyses of variance (ANOVAs) will be run for the set of dependent variables (i.e. BMI, dysfunctional thinking, scores for questionnaires on eating styles, eating disorder pathology, general psychological symptomatology, and self-esteem). Long-term effects of the Think Slim intervention on the set of dependent variables will be examined using one-way repeated-measures ANOVAs collapsing over group (i.e. for all 102 participants), with four levels: pre-Think Slim, post-Think Slim, 3-month follow-up and 12-month follow-up. All analyses will be performed according to the intention-to-treat principle [
69] and the completers principle. Lastly, time-lagged multilevel network analyses will be conducted on data obtained during the pre- and post-Intervention Phase EMA-only weeks, to assess changes in patterns of data that result in unhealthy eating [
31,
37]. Changes in network connectivity between the pre- and post-Intervention Phase networks will be compared between both conditions. Of particular interest are changes in network connectivity and density related to unhealthy eating, in terms of graph analysis measures such as indegree, outdegree, betweenness, and centrality of network nodes.