Feeding Behaviors

The study design comprised an age and gender stratified random sample of residents of the city of Leipzig, in the age group of 40 to 79 years. A subset of 400 participants aged 18 to 39 years was also recruited. In total, 10,000 participants were planned as the target population (Fig. 1). Address lists of randomly sampled citizens were provided by the resident’s registration office of the city of Leipzig. Citizens were sent an invitation letter containing an information leaflet about the study, a response form and a postage-paid return envelope. Persons who did not respond within four weeks received a reminder letter. Non-responders were searched in public telephone directories and contacted by phone. Persons who were interested to participate were scheduled for an appointment in the LIFE study centre. As a prerequisite to enrolment, written informed consent was obtained from all participants. The study was approved by the responsible institutional ethics board of the Medical Faculty of the University of Leipzig. The data privacy and safety concept of the study was approved by the responsible data protection officer. Possible incidents during study visits and during travel to the study site were covered by an insurance policy. Participants received a lump sum of 20 EUR per visit to cover their travel expenses. No other financial incentives were paid out. We carried out several public relation activities to stimulate participation rates. The LIFE study centre was located on the medical campus in the centre of the city, which was easy to reach.Fig. 1

Target sample sizes of the LIFE-Adult-Study

In March 2013, we decided to modify the inclusion criteria of the programme with a reduction of the upper age limit to 74 years. We observed that participants aged ≥75 years had difficulties in completing the assessment programme within the set time limit despite high motivation. Furthermore, it became apparent that participation of women in this age group was markedly reduced (about 1/3 lower than men). The most frequently given reason was that women would not leave their diseased and care-needing partners alone at home on three study days. Therefore, as we stopped recruitment of participants ≥75 years, we extended the lower age limit for deep cognition and depression phenotyping to 60 years. This change was put in place in March 2013 after approval by the institutional ethics board.
In a subset of participants we investigated whether body fat distribution is associated with functional traits of the brain (magnetic resonance imaging, MRI) and traits of eating behaviour. To unravel this question a subcohort of 1200 participants aged 18-79 underwent abdominal MRI-scans in addition to brain MRI-scans.

Objective binge-eating episodes, characterized by (1) “feelings of loss of control over eating behavior” and (2) “consumption of objectively large, inappropriate amounts of food” [1 ,37 (link),40 ], were identified from eating episodes reported over the signal-based (1 item: “Was your meal a main meal, snack, or binge?”) and event-based EMA questionnaires (2 items: “Would other people rate the amount of food as excessive under similar circumstances?” and “Did you feel like you are losing control of your eating behavior?”). The signal-based and the 2 event-based items were recoded into a binary variable indicating the occurrence of an objective binge-eating episode (binge-eating episode reported=1, no binge-eating episode reported=0). As the algorithm was supposed to predict future binge-eating episodes, this variable was shifted backward in time by one signal (approximately 2.5 hour).

A focus group of inpatients (11 female adolescents and young adults in treatment for regular binge-eating episodes at the Schoen Clinic Roseneck, Germany) complemented this literature-based approach. It was conducted to tap into antecedents that nomothetic EMA research might have overlooked so far. After an individual written brainstorming session on “triggers and circumstances associated with binge eating,” the inpatients rated the preliminary list of EMA items on relevance to their binge-eating episodes (“happens before/during/after binge eating…”: 1=[almost] never, 3=might or might not, 5=[almost] always). A moderated discussion of the brainstormed and provided items concluded the sessions.
Next, 2 researchers analyzed the rating data and integrated patient-generated items. This led to the following changes: several constructs missing in the preliminary item list were identified and items were added to cover these gaps (eg, eating based on internal opposed to external motivation: “Did you eat on your own accord?”; (not) following a regular meal structure: “How much did you follow a regular meal structure today?”; and restricting specific foods: “Are you restricting on certain foods right now?”).
The focus group participants further rated 27 of the provided items as positively associated with their binge-eating episodes (mean >3.5), 11 items as negatively associated (mean <2.5), and 9 items as unrelated to their binge-eating episodes (mean 2.5-3.5; Multimedia Appendix 2, Table S1). Some items were scored as unrelated (eg, “Right now I feel: tired” and “I engaged in increased levels of sport.”), and items with large SDs (SD >1.00; eg, “Right now I feel: relived,” “Right now I am shopping for groceries.” and “I acted upon my plans regarding my eating behavior.”) were disregarded, merged (eg, “I am in company.” with “I am on my own.”), or exchanged (eg, “I feel strained due to...work / university / school; close social network; wider social network; everyday stressors” with “Do you feel like you can handle all upcoming tasks and problems?”). As the patients expressed concerns over the redundancy of emotional states, 4 more items were disregarded (“Right now I feel: calm/ashamed/guilty/frustrated”). Finally, 4 items regarding eating behaviors such as “resistance to food craving” or “restriction” were rephrased to map more accurately on constructs introduced by the focus group (see Multimedia Appendix 3, Figure S1 for all item iterations)

All participants completed the following study protocol. First, the BN and BED research diagnoses according to DSM-5 [1 ] were determined via telephone using the Eating Disorder Examination interview [37 (link)] and the Structured Clinical Interview for DSM-IV [38 ]. Both interviews were adapted to the diagnostic criteria of the DSM-5 (eg, 1 binge-eating episode per week for 3 months instead of 2 binges per week for 3 months).
The participants were then introduced to the EMA items and logged into the customized smartphone app SmartEater. SmartEater was used during the subsequent EMA phase, in which signal-based EMA questionnaires were inquired up to 84 times per participant (6 signal-contingent prompts per day, in intervals of 2.5 hours for 2 weeks; questionnaires expired 1 hour after the initial prompt). In addition, an event-contingent EMA questionnaire on overeating, loss of control, and binge-eating episodes was accessible. Participants were instructed to fill in this event-contingent questionnaire whenever they felt like they overate or felt a sense of loss of control over food intake or both. The event-contingent questionnaire included questions to differentiate between subjective and objective binge eating and objective overeating (Multimedia Appendix 4, Table S2). EMA items assessing emotions were presented in a randomized order. However, the other items were presented in a fixed order to prevent carryover effects. The participants were able to review and change their answers through a “back” button. Answering all items (except branched items) was mandatory for submission of the questionnaires.
After the EMA phase of 2 weeks, a JITAI phase of 2 weeks started, in which the participants received short intervention suggestions from the app to prevent binge-eating episodes at ideographically predicted high-risk times. Every study stage was accompanied by web-based questionnaires that assessed current eating behavior pathology, demographic data, perceived acceptability, feasibility, and so on. Data from the intervention phase were not covered in the present article. For reimbursement, the participants received €30 (US $32.80) and personalized feedback on their EMA data and psychometric web-based questionnaires.