Personalising the timing of ecological momentary assessments to assess dietary intake in young adults: a double-blinded crossover formative trial

Ecological momentary assessment (EMA) may be a valid and acceptable method of assessing dietary intake in young adults (1) . EMA may overcome some of the limitations associated with traditional dietary assessment methods such as high respondent burden and memory biases (2) by capturing time-sensitive data via concise dietary surveys. However, most dietary EMA studies either deliver signal-contingent EMAs at fixed intervals or rely on the user’s memory to self-initiate event-contingent EMAs whenever they ate. This may be inappropriate for young adults due to their highly variable eating patterns (1) . Young adults are particularly vulnerable to weight gain due to major life transitions and, for this population, dietary information may need to be collected near real-time to improve recall accuracy (3) . Therefore, the aim of this study was to examine the feasibility (response rate) and acceptability of an EMA protocol that delivered dietary surveys at times personalised to young adults’ (18–30 years) eating patterns and to compare this to the feasibility and acceptability of EMAs delivered at fixed intervals. A randomised, double-blinded crossover design with two four-day treatment arms was used. In one arm, participants received six EMAs per day at fixed intervals. In the other arm, EMAs were delivered at times tailored to participants’ usual eating schedules (ranged between two to six EMAs per day). Usual eating schedules were determined using time-stamped food and beverage images captured by participants over the four days immediately prior to treatments. EMA questions included, but were not limited to, time of consumption and type of food or beverage group consumed. Response rates were calculated as the percentage of EMAs responded to out of the EMAs delivered. At the end of each arm, participants completed an acceptability survey assessing their opinion of the number of EMAs per day, length of the EMAs, and number of recording days. Twenty-three subjects were included (13 female; mean age 26, SD 2.1 years). Mean response rates of the fixed interval and personalised schedule treatments were 65.1% (SE 3.7%) and 66.3% (SE 3.7%), respectively. Compared to the fixed interval treatment, EMAs delivered during the personalised schedule treatment did not align closer with participants’ eating times; the average time difference between EMA delivery and reported eating time was 1.7 hours for both treatments. Participants from both treatments reported receiving too many EMAs per day but found the length of the EMA and number of days of recording to be ‘just right’. In conclusion, EMAs delivered on a personalised schedule may not improve participant adherence. Due to the irregular nature of young adults’ eating patterns, timing of EMA delivery is difficult to tailor. Future definitive trials should use more sophisticated methods of personalisation such as wearable sensors to trigger event-contingent EMAs.