Sage Journals: Discover world-class research

Abstract

Objective

There is limited evidence on the effects of various types of intervention messages on achievement of daily behavioral goals in mobile weight loss programs. Nudge, a 12-week pilot microrandomized trial among young adults, tested whether 7 types of intervention messages based on behavior change techniques (BCTs) increased the likelihood of daily goal achievement.

Methods

Participants received lessons, tailored feedback, self-monitoring tools, and daily messages, and had daily behavioral goals (weighing, active minutes goal, and red foods limit). Four times/day, participants (N = 52, aged 18–35, body mass index 25–40 kg/m²) were randomized to receive or not receive 1 of 7 types of BCT-based messages. Generalized estimating equations evaluated the effects of receiving a message versus not on daily goal achievement, and of receiving and viewing a message versus not viewing a message.

Results

Randomization to receive any message (vs. none) did not impact likelihood of meeting daily goals overall but was associated with fewer red foods tracked (odds ratio (OR) = .96; 95% confidence interval (CI): .94–.98). Randomization to receive (vs. not receive) a social comparison message about red foods was associated with fewer red foods tracked (OR = .92; 95% CI: .85–.99); no other message types were associated with daily goal achievement. Receiving and viewing any message (vs. not viewing) increased odds of achieving daily goals for all 3 behaviors: weighing (OR = 1.91; 95% CI: 1.06–3.43), active minutes (OR = 1.63; 95% CI:1.20–2.20), and red foods limit (OR = 2.09; 95% CI:1.46–3.01).

Conclusions

BCT-based messages may nudge achievement of daily goals when participants view them. Further research should explore how to increase and maintain engagement in mobile behavioral interventions.

Trial registration

Precision Public Health: Enhancing Connections to Develop Just-in-Time Adaptive Intervention Strategies (Nudge); https://clinicaltrials.gov/study/NCT03836391; ClinicalTrials.gov ID: NCT03836391

Keywords

Microrandomized trial just-in-time adaptive intervention mHealth weight goals

Introduction

Obesity remains a public health concern, given that approximately half of adults are expected to have obesity (i.e. body mass index ≥30) by 2030,¹ and obesity is associated with greater risk of comorbidities and mortality.² Young adulthood is characterized by high risk of weight gain, with weight gain of 1 to 2 pounds on average per year from age 18 to 35.^3,4 An estimated 40% of young adults have obesity,⁵ and weight gain during young adulthood is associated with increased risk of cardiovascular disease, type 2 diabetes, hypertension, some cancers, and mortality.^6–8 As promoting weight management to prevent and reduce obesity has the potential to reduce risk of chronic disease later in life, there is a need for interventions to reduce obesity and weight gain in young adults.^6,9

While there is evidence that intensive behavioral modification programs that promote changes in dietary intake and physical activity can lead to clinically relevant weight loss among adults,^10,11 young adults have been largely underrepresented in these trials.^12–15 Given that 97% of young adults between the ages of 18 and 29 own a smartphone,¹⁶ using smartphones or other mobile methods to deliver nutrition, physical activity, and weight management programs can potentially increase the reach and impact to this population. The majority of young adults are interested in programs with reduced face-to-face contact and increased flexibility,¹⁷ and there is evidence that internet- and mobile-delivered programs are feasible, acceptable, and potentially scalable solutions for weight loss.^18–21

Until recently, most in-person and digital programs have used predetermined schedules of content delivery for various standard behavioral components (e.g. lessons, text messages, and tailored feedback) at fixed time points throughout the intervention. Text messages or other brief intervention messages have been used to deliver behavioral strategies that help support behavior change, but the incorporation of all treatment components into a treatment “package” (i.e. a “black box”) means that little has been learned about what types of messages are most effective, or when they should be sent. Digital health tools such as smart scales, activity trackers, and smartphone-based dietary tracking now allow for the continuous collection of daily or within-day self-monitoring data on weight, activity, and dietary intake. The ability to sync and evaluate participant data in real-time has unique advantages in both intervention design and evaluation. Interventions can now be developed that are delivered “just-in-time” that provide highly adaptive, personalized support wherever and only whenever needed, and adapt over time to an individual's changing behavior and context.^22–24 These interventions, called just-in-time adaptive interventions (JITAIs), have shown greater effectiveness relative to comparison and control groups,²⁵ and have focused on behaviors like increasing daily physical activity,²⁶ reducing sedentary behavior,²⁷ and smoking cessation.²⁸ However, most JITAIs to date have focused on a single behavior and have been short in duration. Few JITAIs have used adaptive, tailored treatments specifically for improving dietary behaviors or multiple behaviors needed for weight loss.^29–31

The first JITAI focused on dietary behaviors was tested in an 8-week study in conjunction with a commercially available weight loss website, with the goal of collecting behavioral and contextual data to predict dietary lapses and subsequently deliver tailored intervention messaging to prevent lapses from occurring.³² Another 8-week study among adults with hypertension promoted reduction in sodium intake by delivering personalized messages when a participant was at a restaurant, grocery store, or at home.³³ Though JITAIs are a potentially transformative and scalable approach to delivering obesity interventions, their development is challenged by limited empirical and practical evidence to inform what type, for whom, and under what conditions JIT support is most effective for improving adherence to multiple weight loss behaviors.^34,35

There is little experimental evidence on what types of tailored intervention messages are most effective for promoting daily behavior changes as mediators of weight loss. The microrandomized trial (MRT) is a study design that enables randomization of participants to different interventions one or multiple times per day to test the effect of delivering a single treatment component on the immediate, or proximal, outcome.³⁶ MRTs thus far have evaluated the effect of intervention messaging on physical activity,^26,37–39 dietary adherence,⁴⁰ stress management,^41,42 smoking cessation,^42,43 and app engagement.^44–48 Many behavioral weight loss interventions have created messaging content based on behavior change techniques (BCTs), which are observable and replicable components of interventions designed to bring about change in behavior (e.g. feedback, action planning, and social comparison).^49–51 However, it is not yet known if brief smartphone-based messages based on specific BCTs are effective at promoting adherence to daily self-weighing and activity and dietary goals.

The objective of this pilot study was to test the preliminary effects of BCT-based intervention messages on adherence to daily weight-related behaviors to inform implementation of a fully powered trial. Based on empirical evidence on effective BCTs to promote weight-related behaviors^52,53 and BCTs frequently used in behavioral weight management interventions for young adults,⁵¹ we selected 7 different candidate BCTs to test: (reminders 7.1; outcome feedback, 2.7; social comparison, 6.2; past success, 15.3; goal progress feedback, 1.6; reinforcement, 10.4; implementation intentions, 1.4). An MRT design was used to test which BCT-based intervention messages are effective at promoting achievement of daily weight-related goals (self-weighing, meeting an activity goal, and meeting a dietary goal). We hypothesized that receiving versus not receiving each of the message types would increase likelihood of goal achievement. An additional aim was to examine the relationships between total message views and overall daily goal achievement in the program on weight loss at 12 weeks.

Methods

Study design and participants

This study was a 12-week pilot MRT of BCT-based intervention components designed to promote adherence to daily weight-related behavioral goals. Participants received a weight loss intervention delivered through a smartphone application (i.e. Nudge app, a native iOS application developed for this study), with integrated data from digital devices (wireless scale, Fitbit activity tracker, smartphone, and Nudge app food log), and included several features described below. To test the effects of different intervention components (i.e. 7 BCT-based intervention messages, hereafter called intervention messages) on proximal daily behavioral outcomes, at 4 different times per day (decision points) over 12 weeks, we randomized participants who were available at that time to receive or not receive one of the 7 intervention messages. The microrandomization (participant either received or did not receive the message) allowed each participant to serve as their own control throughout the study.

Young adults were recruited using informational emails sent out through a university listserv and paid Facebook advertisements which directed interested individuals to an online survey to screen for eligibility. Study staff followed up with eligible individuals by phone to describe study details and schedule individuals interested in enrolling for a group session. Participants met the following inclusion criteria: current age 18 to 35, body mass index (BMI) of 25 to 40 kg/m2, self-reported <150 minutes/week of moderate-to-vigorous intensity physical activity (MVPA), English-speaking and writing, owned an iPhone with iOS 11.0 or higher, had internet access and a text messaging plan, and had home wireless internet access compatible with the study wireless scale. Exclusion criteria included: current participation in another physical activity or weight control program, currently pregnant, pregnant in the previous 6 months, or planning pregnancy in the next 3 months, working primarily during night shifts, and health conditions that precluded changes in exercise or diet. Physician consent was required for individuals who reported a diagnosis of or treatment for high blood pressure, high cholesterol, or diabetes. All study procedures were reviewed and approved by the Institutional Review Board of the University of North Carolina at Chapel Hill. Study procedures and data collection took place from February to September 2019; all participants provided informed consent prior to study participation.

Nudge intervention

The 12-week intervention was developed to promote achievement of three daily weight-related behavioral goals detailed below. Participants were given a wireless scale (Fitbit Aria, San Francisco, CA) and physical activity tracker (Fitbit Alta), and downloaded the Nudge smartphone application. Weight and activity data were automatically uploaded to study servers through the Fitbit application programming interface (API). Participants tracked their foods and beverages in the Nudge app food log. To encourage adherence to dietary monitoring, the study used a simplified form of tracking based on the Traffic Light diet that categorizes foods as green (low-calorie, high nutrients), yellow (moderate calories, high nutrients), or red (high-calorie, high-fat, and low nutrients).⁵⁴ This simplified self-monitoring approach has been shown to be feasible, and has resulted in equivalent or higher self-monitoring rates and equivalent weight losses compared to detailed calorie self-monitoring.^55,56 Participants were encouraged to track only their red foods (RF), though in week 1 they were asked to track green, yellow, and red so that they could learn more about the foods in each color category. The food log section of the app included lists of red, yellow, and green foods, with serving size information, and participants could track their foods by meal (Breakfast, Lunch, Dinner, and Snack). Participants were encouraged to select the “No Red Food” button when they consumed 0 RFs for a meal. The Nudge app included separate pages for Weighing, Active Minutes, and Red Foods that displayed participants’ progress toward their daily goals (Figure 1).

Figure 1.

Example Nudge app pages.

During the group kickoff sessions, an interventionist provided a study overview and discussed tracking and daily behavioral goals related to weighing, activity, and eating. Participants were encouraged to focus on limiting their RFs and meeting a personalized daily goal for active minutes to gradually increase activity. The interventionist discussed the rationale for weighing daily and encouraged participants to consider their daily weight as an indicator to help them self-regulate their behaviors (i.e. behavioral changes were working or additional changes were needed). Participants received guidance on sections of the Nudge app, setting up their Fitbit and iPhone settings to ensure frequent syncing of their activity data (i.e. all-day sync, Bluetooth on, background app refresh on, keeping Fitbit app open, and running in background), setting up their scale at home, and how to use the food log. Interventionists discussed the simplified color-coded traffic light system for categorizing foods, gave instructions on logging foods in the app, and provided experiential learning with participants tracking the foods and beverages they consumed the day prior. Participants also received several handouts (food list categories, action steps following session, exercise safety, and setting up the wireless scale).

Daily goals

Participants were encouraged to meet three daily goals: (1) weigh themselves, (2) reach their active minutes (AM) goal, and (3) stay at or under their daily RF limit. Participants’ starting daily AM goal was determined by their self-reported weekly minutes of MVPA assessed at baseline (< 50 minutes/week MVPA = goal of 10 minutes/day; 50-75 minutes/week = 15 minutes/day; 76–100 minutes/week = 20 minutes/day; 101–125 minutes/week = 25 minutes/day; 126–149 minutes/week = 30 minutes/day). Participants used their Fitbit to track AM, which are equivalent to minutes of MVPA that are accumulated in at least 10-minute bouts. Daily AM goals stayed consistent during each program week (Monday–Sunday) but increased by 5 minutes the following week if the participant had reached total weekly AM that were equal to at least 5 times their daily goal.

Participants’ starting daily RF limit was set according to their baseline weight and was designed to produce energy deficits of approximately 300 to 500 calories per day (< 200 lb = 3 RF/day; 200–249 lb = 4 RF/day; ≥ 250 lb = 5 RF/day). The Nudge app did not change RF limits, but the Nudge system alerted study staff of participants who were tracking their RF regularly and staying at or under their RF limit but not losing weight. This circumstance suggested that the RF limit was not low enough to lead to weight loss for the participant, and after review with the study team, study staff would decrease the participant's daily RF limit by 1 (with a minimum limit of 2 RF/day).

Additional program features

The app included 12 lessons, with a new lesson unlocked each week. Lesson content focused on cognitive and behavioral skills such as self-monitoring, problem solving, overcoming barriers, social support, goal setting, cognitive restructuring, and making peace with the scale (i.e. seeing weight as an objective indicator of how daily diet and activity choices affect weight, rather than as a judgment). Resources included information specific to using the Nudge eating plan, food lists categorized by green, yellow, and RF, how to use a Fitbit, and a video about the benefits of daily weighing. At the beginning of each week, participants also received personalized feedback about their progress overall and in the past week. Feedback was tailored using computer algorithms that provided comprehensive feedback on participants’ weighing, AM, and RF intake in the context of their overall and recent weight change.

Intervention messages based on behavior change techniques

Within the Nudge app, we tested 7 different types of intervention messages (Table 1) designed to deliver distinct BCTs, which are observable and replicable components of interventions that are designed to bring about behavior change.⁵⁰ We selected BCTs that are commonly used in behavioral weight loss interventions with young adults,⁵⁷ have demonstrated efficacy for improving behavioral mediators of weight loss,^52,53 or have the potential to promote overall self-regulation but have not been systematically evaluated for their effects on weight-related behaviors. We hypothesized that receiving, compared to not receiving, each of these 7 types of intervention messages would promote proximal change (either on the day of or day after receiving a message) by increasing the likelihood of participant achievement of the weight-related behavioral goals. The 7 types of intervention messages included: (1) reminders for weighing and self-monitoring of AM and RF (reminders; 7.1); (2) feedback on outcome of behaviors (i.e. weight change since last weighing) (outcome feedback; 2.7); (3) averaged data about other participants’ behavioral performance from the previous day or previous week (social comparison; 6.2); (4) data on the participant's past success in performing a behavior, when available (past success; 15.3); (5) data on the participant's progress toward the goal that day (i.e. discrepancy between current behavior and daily goal) (goal progress feedback; 1.6); (6) positive, reinforcing statements about goal achievement (reinforcement; 10.4); and (7) prompts for action planning or implementation intentions (“if/then” plans) related to a behavior (implementation intentions; 1.4). Message types 1 to 6 were static messages; implementation intentions were interactive messages with text boxes for the participant to enter their plans. Messages for the seven intervention types were developed separately for each of the three behaviors (weighing, AM, and RFs), and each intervention message focused on a single behavior (Table 1).

Table 1.

Examples of Nudge intervention message types.

Decision points	Primary outcomes [secondary outcomes]	General decision rules for availability	Example message
Reminders
Early morning	Weigh today	Did not weigh yesterdayNot out of town	Get off to a great start this morning by stepping on the scale. Small decisions add up over time to help you reach your goals!
	Wear tracker today	Did not wear tracker yesterday	Remember to wear your Fitbit today. We use your Fitbit data to give you personalized support that will help you meet your active minutes goals every day
	Track RF today	Did not track RF yesterday	Keeping track of your food is one of the most important tools you can use to lose weight. Get back to tracking your red foods today!
Outcome feedback
Late morning	Weigh today^a	Weighed today or yesterdayNot out of town	Your weight today was 0.7 lbs above where it was the last time you weighed. Weighing every day will keep you focused on your goals and will help you tip the scale the other way
	Met AM goal today^b[total AM today]	Weighed today or yesterday	Your weight today was 0.4 lbs below where it was the last time you weighed. Keep up the hard work and make choices that align with meeting or exceeding your active minutes goal!
	Met RF limit today^c[total RF today]	Weighed today or yesterday	Your weight yesterday was 1.2 lbs below where it was the last time you weighed. Props to you! Keep your eye on the prize and stick to your red food limit to meet your goals
Social comparison
Late morning, afternoon	Weigh today^a	Not out of town	Yesterday, 55% of participants stepped on the scale. If you haven't already, boost this percentage by stepping on the scale before the end of the day
	Met AM goal today^b[total AM today]	None	Last week, participants had an average of 140 total active minutes by the end of the week. Can you top that? Start today by getting on your feet!
	Met RF limit today^c[total RF today]	None	Participants in this study stayed below their red food limit an average of 3.4 days last week. Can you beat that? Start by eliminating a common red food in your diet today
Past success
Late morning, afternoon	Weigh today	Not weighed today or yesterday Prior to this week, had ≥ 1 week of weighing on ≥ 3 days Not out of town	In week 3 you weighed yourself 6 days. You have the skills to meet your goals. Keep your scale in plain sight to make daily weighing a no-brainer. Step on the scale today!
	Met AM goal today	Not yet met AM goal today Not met AM goal for ≥ 2 out of last 3 days Prior to this week, had ≥ 1 week of meeting AM goal on ≥ 3 days	In week 2 you met your active minutes goal on 5 days. Keep your health a top priority! This week, try a new exercise routine to keep things interesting, maybe a hike or a workout on YouTube
	Met RF limit today	Not yet exceeded RF limit today Exceeded RF limit or did not track for ≥ 2 out of last 3 days Prior to this week, had ≥ 1 week of meeting RF limit on ≥ 3 days	In week 2 you met your red food limit on 6 days. Today is another opportunity to make it happen. Can you start again today?
Goal progress feedback
Afternoon	Weigh today	Not weighed yet today Not out of town	Weighing every day is best for optimal weight loss. You haven't weighed yet today. Take a moment to reflect on how this daily behavior has changed your awareness about your eating and activity
	Met AM goal today^b[total AM today]	None	Today you've had 14 active minutes out of your goal of 20 minutes. Head out for a walk after dinner tonight to meet or exceed your goal
	Met RF limit today^c[total RF today]	None	So far today you've tracked 2 red foods out of your limit of 3. If you feel tempted to snack today, try grabbing a green food!
Reinforcement
Evening	Weigh tomorrow	Weighed today	Thank yourself for stepping on the scale today. You are building a lifestyle for a healthy future. Stick with it!
	Met AM goal tomorrow	Met AM goal today	Great job meeting your active minutes goal today! Keep up the good work. Staying active will help you live a long and healthy life
	Met RF limit tomorrow	Met RF limit today (tracked complete day and stayed at or under RF limit)	Terrific job staying under your red food limit today! A habit of healthy eating will help you reach your weight goals
Implementation intentions
Evening	Weigh tomorrow	Did not weigh today Did not weigh yesterday Not out of town	To remind myself to weigh tomorrow, I will [enter plan to remind yourself]. If I am tempted not to step on the scale, I plan to [enter plan to deal with temptation]
	Met AM goal tomorrow	Did not meet AM goal today Did not meet AM goal yesterday	Tomorrow, I plan to exercise around [enter time]. To keep me accountable for my exercise, I will tell [enter name of friend or family member] my plan
	Met RF limit tomorrow	Exceeded RF limit or did not track any foods today Exceeded RF limit or did not track complete day yesterday	I plan to eat [enter number] red foods tomorrow. If I stay under my red food limit tomorrow, I will reward myself by [enter nonfood reward plan]

Analysis limited to participants who had not yet weighed at time of randomization.

Analysis limited to participants who had not yet met AM goal at time of randomization.

Analysis limited to participants who had not yet exceeded RF limit at time of randomization.

AM: active minutes; RF: red foods.

Microrandomization of intervention messages

Participant eligibility for messages was evaluated at four prespecified times per day (early morning: 7:00AM, late morning: 10:00AM–12:00PM, afternoon: 2:00–4:00PM, and evening: 7:00–9:00PM) using decision rules that incorporated data regarding weight, AM, and RFs. Table 1 depicts each of the seven types of intervention messages and their relevant time points for delivery, primary outcome(s), decision rules for eligibility, and example messages. All seven intervention messages were not available for randomization at all decision points. At each of the four randomization decision points, the system evaluated the sets of decision rules for each intervention message available for randomization during that time period to determine which messages the participant was eligible to receive. Participants could only receive a maximum of three messages per day, one per behavior, such that the behavior that was the focus of the first randomization, whether the message was received or not received, was eliminated from message options for later time periods, and similarly for the second randomization. After evaluating the decision rules and determining which intervention message(s) the participant was eligible to receive (if any), the system would randomly select one from the available message types. At that point, the system microrandomized the participant with equal (50%) probability to receive or not receive the message. When randomized to receive a message, participants received a push notification that said, “Hello, check out a new message in Nudge!” as well as auditory alerts. The message was displayed in the app on the relevant behavior page (i.e. message about weighing displayed at top of Weighing page). If a message had been received but not yet viewed, both the smartphone icon for the Nudge app and an icon on the home page of the Nudge app displayed a red badge notification as additional cues to the participant that a new message was available. The messages were available until the end of the day and at 12:00AM they disappeared.

Measures

Proximal outcomes

Daily weighing. Weight and weighing behavior were measured daily at the participant's home using the Fitbit smart scale, which transmitted weights directly to study servers and has been shown to have strong concordance with a research-grade scale.^58,59 Participants were instructed to weigh themselves each day in light clothing, without shoes. Weigh today was calculated as a dichotomous indicator of weighing or not weighing on the day of randomization to receive or not receive a message; weigh tomorrow was calculated from weighing on the day after randomization.

Active minutes and tracker wear. Fitbit trackers, which have demonstrated validity for measuring time in physical activity when compared with research-grade accelerometers,^60,61 were used to collect data on daily AM and steps on the day of randomization. Each time the participant synced their tracker with the Fitbit app on their phone, data were transmitted to the Fitbit API and then to study servers. Wear tracker today was calculated as a dichotomous outcome of whether a participant wore a tracker or not (i.e. tracker wear was defined as ≥1000 steps by end of day on the day of randomization.⁶² Met AM goal today and met AM goal tomorrow were calculated as dichotomous outcomes of whether a participant met or exceeded their daily AM goal or not (days with no tracker wear were counted as not meeting goal) on the day of randomization, or day following randomization, respectively. Total AM today was calculated as the accumulated number of AM on the day of randomization.

RFs and RF tracking. RFs were collected from data participants recorded in the food log of the Nudge app, which synced with study servers in real time. Track RF today was calculated as a dichotomous indicator of a complete or incomplete day of RF tracking on the day of randomization. A complete day of RF tracking was defined as RFs tracked, or selected the “No Red Food” button, during breakfast or lunch and dinner, or reached/exceeded their RF limit. Met RF limit today and met RF limit tomorrow were calculated as dichotomous measures of whether a participant stayed at or under their RF limit or not on the day of, or day following, randomization, respectively. Days without complete RF tracking were considered not meeting the RF limit. Total RF today was calculated as the total number of RFs tracked on any given day of randomization with complete RF tracking.

Message receipt and views. Messages were categorized as received versus not received when a participant was randomized at a given decision point. This was defined only by randomization; participants did not have to view the message to count as having received the message. Messages received were also categorized as viewed versus not viewed using app engagement data. A message was counted as viewed if the participant viewed the page on which the message was placed (Weighing, Activity, or Eating) any time after it was delivered and before midnight on the day of randomization. Total message views was a sum of all messages the participants viewed in the 12-week study.

Total goals met. A total goal score for the three primary behavioral goals was calculated by summing the total days weighed, total days met AM goal, and total days met RF limit across the 12-week study (range of 0–252).

Other measures

Demographic characteristics, including age, race, gender, education level, and income, were assessed at baseline using online REDCap questionnaires used in our previous studies among young adults.^63,64 Height was measured in person at baseline using a wall mounted stadiometer, with two measures taken and averaged. Weight was measured at baseline and 3 months in person on calibrated scales with participants in light clothing, without shoes. Two measures were administered and averaged. Height and weight measures were used to calculate BMI (weight in kilograms divided by height in meters squared).

Statistical analyses

Given that this was a pilot study, it was not designed to power for all comparisons planned in this paper; instead, it was powered to detect an effect of being randomized to receive any message versus no message. Power analyses for the effect of receiving any message, versus no message, on goal achievement were conducted with the MRT sample size calculator for 80% power to detect an average proximal constant treatment effect (standardized; proximal effects/average standard error) of 0.08, assuming an alpha of .05 and an average of 30% availability.^65–67

All analyses were conducted using RStudio statistical software (Version 4.2.1). Descriptive statistics for demographic variables were calculated for the sample. Our primary analyses were to evaluate the effects of: (1) any intervention message (vs. randomized to not receive a message) on adherence to daily weight-related behaviors on the day of randomization (weigh today, met AM goal today, total AM today, met RF limit today, total RF today); (2) behavior-specific intervention messages (vs. randomized to not receive a message) on adherence to the relevant daily behaviors on the day of randomization (weigh today, met AM goal today, total AM today, met RF limit today, total RF today); and (3) the unique effects of the seven intervention message types for each behavior (BCT-based messages) on adherence to their relevant daily behaviors on the day of or day following randomization. This was done using generalized estimating equations with an independence working correlation matrix and robust standard errors for this analysis to account for repeated measures within individuals. The use of random effects has been discouraged within the context of MRTs because they muddy our ability to make causal inference.⁶⁸ We specified a logit link for binary outcomes and a log link for count outcomes.

For each outcome of interest, the analytic sample was limited to time points when participants were determined to be eligible and were randomized to receive or not receive an assigned intervention message. Table 1 specifies the outcomes for each intervention message type. Outcomes varied as some message types were delivered more frequently than others (i.e. delivered whether or not participants were meeting their goal at the time (outcome feedback, social comparison, and goal progress feedback), delivered only when participants were not meeting their goal (past success), and delivered only in the evening and were hypothesized to affect the next day's behavior (reinforcement and implementation intentions). For analysis of dichotomous outcomes (weigh, met AM goal, and met RF limit), the eligible sample was limited to observations in which participants had not already met their daily goal at the time of randomization to evaluate the effect of the message on meeting the specific goal (e.g. not met AM goal). For analysis of continuous outcomes, all randomization observations were included. Results are presented as untransformed coefficients. For dichotomous outcomes, exponentiating coefficients corresponds to odds ratios. For count outcomes, exponentiating coefficients correspond to rate ratios (i.e., multiplicative factors). We applied the Benjamini-Hochberg correction⁶⁹ to retain a 5% false discovery rate across the multiple tests, using a logit link for binary outcomes, a log link for total AM today, and an identity link for total RF today.

In exploratory analyses, we examined the effect of receiving and viewing versus receiving and not viewing messages (any message, behavior-specific messages, and BCT-based messages) on achievement of daily goals. We excluded observations for which no message was received. In line with similar studies that used a stabilized weighting approach to account for time-varying confounders in MRT studies,^39,66,67,70 we weighted observations for which a message was randomized to be delivered by the inverse of the probability of the message action taken and by message availability and centering the treatment effects for added robustness. Inverse probability weighting adjusts for time-varying confounding of the exposure effect given individuals’ past behavior. This approach is similar to propensity score weighting in that observations in which actions were less likely to have occurred are given more weight than observations that were likely to have occurred.

To create the stabilized inverse probability weights (SIPWs), the numerator was each person's historical probability of viewing a message that they received in the program so far (i.e. up to the point of randomization), and the denominator was probability of viewing the current message, which was calculated with an equation using time of day and recent goal achievement in the past 7 days (as these variables were associated with probability of viewing a message). The probability of achieving a daily goal after viewing a message was estimated using logistic regression models that controlled for previously identified variables associated with message views, including history of message viewing, length of time in program, time of day, total weight change, weight change in last week, recent weighing lapse (vs. none), and dietary goal achievement in past week.⁷¹ In sensitivity analyses comparing results with and without SIPWs, effects were uniformly smaller when accounting for SIPWs, so we report findings when controlling for SIPWs.

To determine if daily goal achievement was associated with weight loss at the end of the program, separate analysis of variance models evaluated the effect of total number of days each behavioral goal was met (weigh, wear tracker, met AM goal, track RF, and met RF limit), and total number of days the three primary daily goals were met (weigh, met AM goal, and met RF limit) on 12-week weight change, controlling for baseline weight.

Results

Participants

A total of 53 young adults were enrolled in the study, the majority of whom were female (79%) with a college or higher degree (85%). Mean BMI was 31.9 kg/m² (SD = 4.4), and over one-third of the sample identified as Black or from other racial/ethnic minority backgrounds (Table 2). One participant withdrew on the third day of the study due to pregnancy and never received any messages; thus, the full sample consisted of 52 participants. Two additional participants withdrew from the study at days 43 and 60, reporting that they disliked intervention messages and had unforeseen circumstances respectively; their message and behavioral data are included up until the point of withdrawal (Figure 2, CONSORT).

Figure 2.

CONSORT diagram.

Table 2.

Baseline demographic characteristics (n = 53).

Variable	% (n)/Mean(SD)
Gender
Female	79.2% (42)
Male	20.8% (11)
Age	29.6 (3.8)
BMI	31.9 (4.4)
Race
White	62.3% (33)
Black	16.9% (9)
Other or multiple races*	20.8% (11)
Education
High school grad or GED	3.8% (2)
Some college or associate degree	11.3% (6)
College graduate	47.2% (25)
Masters or doctoral degree	37.7% (20)
Annual income
<$5000–$15,999	5.7% (3)
$16,000–$24,999	7.5% (4)
$25,000–$34,999	9.4% (5)
$35,000–$49,999	13.2% (7)
$50,000–$74,999	32.1% (17)
$75,000–$99,999	11.3% (6)
$100,000+	17.0% (9)
Don’t Know	3.8% (2)

*Other races include: American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, or reported “Other” SD: standard deviation; BMI: body mass index

Decision points and intervention messages delivered

Of 16,086 decision points when participants were eligible for randomization, 127 observations were removed due to system errors that resulted in 2 or 3 randomizations that occurred simultaneously, and it was not possible to determine what the participant could see on their app during those times. Participants were not eligible for an intervention during an additional 3434 decision points (i.e. 21.3% of the time). Thus, 12,525 decision points were available for analysis. Across the decision points when participants were eligible for randomization, participants were randomized to receive 6183 messages (i.e. 49% of the time), with 18% focused on weighing, 39% on AM, and 43% focused on RF (Table 3). The percentage of messages delivered across the 7 intervention types ranged from 8% to 27%. We excluded an additional 630 observations from models that evaluated weighing as an outcome because of an error in variable coding that indicated whether participants were out of town or not, which affected availability for messages related to weighing for a short time period.

Table 3.

Total number and percentage of each intervention message type delivered across 12-week study.

Message type	Weighingn	Active minutesn	Red foodsn	Total of message typen (% of total)
Reminders	81	218	780	1079 (17.4)
Outcome feedback	338	309	256	903 (14.6)
Social comparison	232	798	637	1667 (27.0)
Past success	50	297	325	672 (10.9)
Goal progress feedback	33	343	291	667 (10.8)
Implementation intentions	51	379	288	718 (11.6)
Reinforcement	307	57	113	477 (7.7)
Total of behavior-specific messages	1092	2401	2690	6183

Effects of receiving (vs. not receiving) any intervention messages

Table 4 shows the estimated effects of messages on adherence to daily weight-related behaviors on the day of, or day following, randomization. On average, over the types of intervention messages and study days, the effect of receiving any message versus not receiving a message was associated with a lower number of RFs (odds ratio (OR) = .96, 95% confidence interval (CI): .94, .98, p = .007) on the day of randomization. Receiving any message versus not receiving a message was not associated with odds of weighing, meeting AM goal, meeting RF limit, or total AM on the day of randomization.

Table 4.

Odds ratios and incident rate ratios comparing effect of randomization to receive versus not receive messages on daily behavioral outcomes.

	Wear trackerOR (95% CI)n	Track RFOR (95% CI)n	WeighOR (95% CI)n	Met AM goalOR (95% CI)n	Total AM goalIRR (95% RR)n	Met RF limitOR (95% CI)n	Total RFIRR (95% RR)n
Any message^a	-	-	0.89 (0.77, 1.02)2035	0.99 (0.92, 1.07)11,152	0.99 (0.98, 1.06)12,526	1.03 (0.95, 1.11)12,304	0.96 (0.94, 0.98)^†7721
Behavior-specific message^b	-	-	1.03 (0.72, 1.47)440	1.00 (0.92, 1.21)4358	0.99 (0.92, 1.07)4838	0.97 (0.88, 1.07)5349	0.97 (0.93, 1.01)3392
BCT-based and behavior-specific message^c
Reminders	0.70 (0.48, 1.01)422	1.02 (0.87, 1.19)1627	1.01 (0.53, 1.93)120	-	-	-	-
Outcome feedback	-	-	1.47 (0.28, 7.59)45	1.08 (0.81, 1.45)590	0.98 (0.85, 1.12)642	0.76 (0.59, 1.07)516	1.00 (0.87, 1.15)398
Social comparison	-	-	1.05 (0.54, 2.05)99	1.00 (0.82, 1.22)1482	0.97 (0.88, 1.07)1653	0.89 (0.70, 1.12)1289	0.92 (0.85, 0.99)^†865
Past success	-	-	0.77 (0.28, 2.14)66	1.13 (0.79, 1.60)588	-	0.99 (0.74, 1.33)660	-
Goal progress feedback	-	-	1.94 (0.51, 8.26)49	0.96 (0.64, 1.41)546	1.06 (0.91, 1.24)679	1.15 (0.81, 1.64)522	0.90 (0.79, 1.04)373
Implementation intentions		-	1.33 (0.43, 4.16)*61	1.27 (0.89, 1.81)*730	-	0.47 (0.22, 1.01)*124	-
Reinforcement	-	-	0.70 (0.44, 1.10)*523	0.97 (0.56, 1.68)*124	-	1.01 (0.58, 1.75)*213	-

Results are provided as OR (95% CIs) for categorical outcomes and IRR (95% CIs) for continuous outcomes. Bolded estimates indicate CIs that did not include 1.0.

Effect of receiving any message on the behavioral outcome.

Effect of receiving any behavior-specific message on the behavioral outcome.

Effect of receiving behavior-change-technique- and-behavior-specific message on the behavioral outcome.

†

No longer significant at the p < .05 level after the Benjamini-Hochberg correction.

*Effect of message on the behavioral outcome tomorrow (i.e. day following randomization).

AM: active minutes; BCT: behavior change technique; CI: confidence interval; IRR: incident rate ratios; OR: odds ratios; RF: red foods.

Effects of receiving (vs. not receiving) behavior-specific messages

When evaluating the effects across the intervention message types focused on a specific behavior (i.e. weighing, AM, or RF) and across study days, on average, receiving behavior-specific messages did not have any effects compared with not receiving them (Table 4).

Effects of receiving (vs. not receiving) BCT-based messages

When examining the unique effects of the seven BCT-based behavior-specific intervention types on their relevant behavioral outcomes (Table 4), only social comparison messages about RFs had a positive effect, such that receiving a social comparison message versus not receiving one was associated with reduced number of RFs tracked (incident rate ratio = .92, 95% CI: .85, .99). Receiving a reminder message, versus not receiving one, did not have effects on any of the intended tracking behaviors on the day of message randomization (i.e. weigh today, wear tracker today, and track RF today). Similarly, receiving messages about outcome feedback, past success, goal progress feedback, implementation intentions, or reinforcement, versus not receiving a message, did not produce any changes in proximal behaviors.

Inverse probability weighted effects of viewing (vs. receiving and not viewing) any intervention messages

Results are shown in the top row of Table 5.When accounting for stabilized inverse probability weights (SIPWs), receiving and viewing any intervention message, versus receiving and not viewing messages, was associated with higher odds of weighing (OR = 1.91, 95% CI: 1.06, 3.43), meeting AM goal (OR = 1.63, 95% CI: 1.20, 2.20), and meeting RF limit (OR = 2.09, 95% CI: 1.46, 3.01) on the same day the message was received, but was not associated with total AM and total RF. These effects remained significant after adjustment for multiple comparisons.

Table 5.

Odds ratios and incident rate ratios comparing effect of viewing messages received versus not viewing messages received on daily behavioral outcomes.

	Wear trackerOR (95% CI)n	Track RFOR (95% CI)n	WeighOR (95% CI)n	Met AM goalOR (95% CI)n	Total AM goalIRR (95% RR)n	Met RF limitOR (95% CI)n	Total RFIRR (95% RR)n
Any message^a			1.91 (1.06, 3.43)1016	1.63 (1.20, 2.20)5521	1.11 (0.90, 1.38)6183	2.09 (1.46, 3.01)6080	0.93 (0.82, 1.06)3807
Behavior-specific message^b			1.04 (0.53,2.06)213	1.54 (1.09, 2.15)2174	1.07 (0.85, 1.34)2401	2.41 (1.41, 4.12)2643	0.94 (0.78, 1.13)1662
BCT-based and behavior-specific message^c
Reminders	1.35 (0.72, 2.54)218	2.70 (1.32, 5.53)^†780	1.08 (0.27, 4.27)51
Outcome feedback			2.04 (0.18, 23.33)19	1.12 (0.50, 2.53)293	1.03 (0.62, 1.71)309	1.70 (0.62, 4.64)253	0.89 (0.66, 1.20)191
Social comparison			0.43 (0.11, 1.75)47	1.37 (0.87, 2.18)715	1.07 (0.81, 1.42)798	2.58 (1.43, 4.64)^†631	0.77 (0.58, 1.03)401
Past success			2.71 (0.88, 8.35)36	1.49 (0.71, 3.11)^†297		4.11 (1.70, 9.90)^†325
Goal progress feedback			0.56 (0.09, 3.35)25	2.00 (0.96, 4.17)272	1.06 (0.77, 1.45)343	6.60 (2.31, 18.87)^†285	0.80 (0.61, 1.07)208
Implementation intentions			4.40 (0.38, 51.1)*35	1.60 (0.85, 3.02)*379		1.09 (0.44, 2.75)*288
Reinforcement			3.44 (1.22, 9.70)^†*285	1.85 (0.35, 9.80)*57		^d113

Results are provided as OR (95% CIs) for categorical outcomes and IRR (95% CIs) for continuous outcomes. Bolded estimates indicate CIs that did not include 1.0.

Effect of receiving and viewing any message on the behavioral outcome.

Effect of receiving and viewing any behavior-specific message on the behavioral outcome.

Effect of receiving and viewing BCT-behavior-specific message on the behavioral outcome.

Near complete separation; cannot estimate model.

†

No longer significant at the p < .05 level after the Benjamini-Hochberg correction.

*Effect of message on the behavioral outcome tomorrow (i.e. day following randomization).

AM: active minutes; BCT: behavior change technique; CI: confidence interval; IRR: incident rate ratios; OR: odds ratios; RF: red foods.

Inverse probability weighted effects of viewing (vs. receiving and not viewing) behavior-specific messages

Results are shown in the second row of Table 5. Viewing messages about activity, versus not viewing them, was associated with higher odds of meeting AM goals (OR = 1.54, 95% CI: 1.09, 2.15) on the day of message receipt. Receiving and viewing messages about RFs, versus not viewing them, was associated with higher odds of meeting a RF limit (OR = 2.41, 95% CI: 1.41, 4.12) on the same day as the message. These effects remained significant after adjustment for multiple comparisons. Viewing behavior-specific messages was not associated with weighing, total AM, or total RFs tracked on the day of message receipt.

Inverse probability weighted effects of viewing (vs. receiving and not viewing) BCT-based messages

Results are shown in the last seven rows of Table 5. Viewing reminder messages for RF tracking was associated with increased odds of completing RF tracking on that day compared to receiving and not viewing them (OR = 2.70, 95% CI: 1.32, 5.53, n.s. after Benjamini-Hochberg correction). Viewing reminder messages for tracker wear and weighing did not impact likelihood of those behaviors on the day of message receipt. Viewing reinforcement messages about weighing was associated with greater odds of weighing the next day (OR = 3.44, 95% CI: 1.22, 9.70), while viewing other BCT-based messages about weighing were not associated with an increased likelihood of weighing that day. Viewing, versus not viewing, BCT-based messages about activity did not affect the likelihood of meeting an AM goal. Meeting a RF limit was more likely after viewing social comparison messages (OR = 2.58, 95% CI: 1.43, 4.64), past success messages (OR = 4.11, 95% CI: 1.70, 9.90), and goal progress feedback messages (OR = 6.60, 95% CI: 2.31, 18.87), compared to not viewing them, but was not significant for other message types. Total AM and total RF were not affected by viewing of any BCT-based activity and RF messages, respectively. None of the BCT-based message effects on weighing or meeting the RF limit remained significant after applying the Benjamini-Hochberg correction for multiple comparisons.

Weight change and behavioral goal achievement

Participants with complete weight measurements at 12 weeks (n = 51) had an average weight change of −2.7 kg (SD = 3.6) and −3.0% (SD = 4.0). Out of a maximum of 84 days, median days weighed was 72 (IQR 64–80), median days meeting the AM goal was 37 (interquartile range (IQR) 19–52), and median days meeting the RF limit was 36 (IQR 16–50).

Associations between message views, goal attainment, and weight loss

Total message views in the 12-week program were significantly associated with weight (kg) lost (B = −0.04, SE = .017, p = .01). The total number of the 3 primary daily behavioral goals met (weighing, meeting AM goal, and meeting RF limit) across the 12-week program was associated with weight loss, such that for each additional goal met, weight loss increased by .05 kg (B = −.050, SE = .009, p < .0001, Table 6). When examining by behavior-specific goals, total days weighed (B = −0.07, SE = .026, p < .01), total days met AM goal ((B = −0.07, SE = .024, p = .003), and total days met RF limit (B = −0.12, SE = .017, p < .0001) were all associated with weight loss at 12 weeks, such that greater days the goals were met was associated with greater weight loss. Similarly, total days of self-monitoring AM (B = −0.09, SE = .033, p = .01) and RFs (B = −0.10, SE = .018, p < .0001) were associated with 12-week weight loss.

Table 6.

Effect of total number of days met behavioral goals on 12-week weight change.

Variable	B*	Standard error	p
Total days wore tracker	−0.09	.033	.01
Total days with complete RF tracking	−0.10	.018	< .0001
Total days weighed	−0.07	.026	< .01
Total days met AM goal	−0.07	.024	.003
Total days met RF limit	−0.12	.017	< .0001
Total behavioral goals met**	−0.05	.009	< .0001

*Bivariate associations.

**Behavioral goals include days weighed, days met AM goal, and days met RF limit (goals range from 0 to 3 per day for a range of 0 to 252 across the 12-week study).

AM: active minutes; RF: red foods.

Discussion

In this MRT we tested the effects of seven different just-in-time adaptive intervention messages on likelihood of adhering to multiple weight-related behavioral goals. Overall, we found that receiving any message may be associated with a small reduction in the total number of RFs consumed, and BCT-based behavior-specific messages may be feasible for promoting daily adherence to self-weighing and staying at or under a RF limit, particularly when participants view the message.

When examining BCT-based messages, proximal effects on their relevant behaviors were only observed for social comparison messages about RFs. All participants were eligible for social comparison messages, whether or not they had exceeded their goal at the time of randomization, so these analyses had more power to evaluate the relationship compared to other message types. However, it is notable that a message indicating how other study participants performed (e.g. % met RF limit yesterday), and asking young adults to consider how they compared or whether they could do better, was associated with fewer total number of RFs consumed when compared to not receiving the message, though it was no longer significant after adjustment for multiple comparisons. While social comparison messages were not associated with meeting the RF limit, the messages appeared to have a positive impact on minimizing the total number of RFs consumed, especially on days when participants had already exceeded their RF limit. These messages were delivered in the late morning and afternoon time points, so perhaps participants had enough time to consider how others were doing in the program and were motivated to make changes to their eating behaviors for the remainder of the day. Mobile apps for weight management and physical activity promotion commonly use social comparison features,⁷² yet there is limited evidence evaluating these features as a mechanism for promoting healthy dietary intake.⁷³ There is evidence that social norms are associated with eating behaviors in young adults.^74,75 Given the high prevalence of social media use among young adults, it is possible that exposure to food-related posts external to the Nudge program may have influenced participants’ perceived norms and food intake. For instance, perceived injunctive norms of other social media users’ intake of energy-dense foods have been shown to be associated with study participants’ own intake of energy-dense foods,⁷⁶ and in a randomized controlled trial, exposure to junk food content on social media impacted young adults’ food choice.⁷⁷ Individuals’ tendency to value comparison to others (i.e. social comparison orientation) has been shown to be associated with weight loss outcomes, but it is unclear whether upward or downward comparisons are more effective for promoting weight-related behaviors.⁷⁸ As participants were eligible to receive social comparison messages regardless of whether they had met the daily behavioral goal or not (i.e. upward or downward comparison), additional research is needed to determine how the effects of these messages varied by individual contexts.

No other BCT-based messages were associated with the daily behavioral outcomes. Possible reasons for this finding include lack of power to detect small effects and the possibility that participants began to expect which messages they would be receiving over time. Though each message library for each BCT-behavior combination included 12 messages (e.g. 12 reinforcement messages for weighing, etc.), the content and timing of these messages did not change during the study. Thus, it is possible that participants learned which message types they would be receiving during certain times of day and were less likely to view them if they were not expecting novel content. Additionally, it is possible that the effects of BCT-based messages on behavioral outcomes may depend on individuals’ motivations to weigh themselves, increase their physical activity, or eat a healthier diet. Previous research has shown motivation to be an underlying mechanism of action through which BCTs exert their effects on behavior change.^79,80 For example, implementation intentions are more effective for behavior change when participants have high motivation to achieve a goal.^81,82 It is possible that the lack of BCT-based message effects on daily goal achievement may be due in part to low autonomous motivation to adhere to the daily goal behaviors. Future work could benefit from examining individual contexts, both psychosocial and behavioral, as moderators of BCT-based message effects.

When exploring the effects of receiving and viewing BCT-based messages, more proximal effects were observed and in expected directions. An important consideration is that participants were not randomized to view or not view the message, and factors related to their program participation thus far may have affected their likelihood of opening the app and viewing a message.⁸³ However, analyses attempted to control for confounding factors by including covariates for recent weight change and goal achievement. Social comparison messages, past success, and goal progress feedback messages, when viewed, appeared to positively impact achievement of meeting RF goals compared to when not viewed. Reinforcement messages positively impacted weighing goals when viewed.

Of the two message types delivered in the evening, when viewed, only reinforcement had a positive effect on the next day's behavior, and only for weighing, such that receiving and viewing a positive reinforcement message about weighing was associated with greater likelihood of weighing the following day compared to not viewing it. Systematic reviews have demonstrated that BCTs of nonspecific reward and social rewards produced positive effects on physical activity among those with obesity⁵³ or cancer,⁸⁴ though that was not found in the current study. However, reinforcement messages for weighing may have had a greater impact than those focused on AM or RF goals because the weighing behavior most often occurred the following morning and was not as susceptible to barriers that may arise throughout the day that make it more difficult to adhere to AM and RF recommendations. It is possible that reinforcing messages were helpful for those who had low motivation to weigh themselves daily. Previous study participants who have been taught to weigh themselves daily in the context of a weight management program have reported that daily weighing is a useful tool to control their weight and increase their awareness of dietary and activity behaviors, while some have indicated weight fluctuations were discouraging and a barrier to daily weighing.⁸⁵ Future work is needed to better personalize messages based on individual perceptions of what young adults would find to be a positive reinforcement with respect to adhering to specific behaviors.

Viewing implementation intention messages had no effect on the next days’ behavioral goal achievement. Prompting implementation intentions has been shown to improve weight loss over short time frames (i.e. 4–10 weeks),^86–88 but these studies examined the effects of implementation intentions as part of a comprehensive program on overall weight loss over time and prompted them less frequently (e.g. once and weekly),^86–88 as opposed to the isolated effect of an implementation intention message on daily proximal weight-related outcomes. Additionally, in this study, implementation intention messages were interactive, such that participants were prompted to fill in text boxes to set their plan for the next day. It is possible that participants viewed the activity but did not complete it, which would preclude an effect of viewing on goal achievement. Given that implementation intentions are more effective when participants have high motivation to change their behavior,⁸¹ low motivation possibly could have influenced participant receptivity and completion of these messages. Among young adults, there is limited evidence on whether frequent implementation intention prompts can promote daily achievement of multiple weight-related behavioral goals.⁸⁷ There is evidence that utilizing action planning together with other BCTs may be more effective for promoting physical activity than action planning alone.⁸⁹ A review of 22 systematic reviews that evaluated the effectiveness of BCTs in promoting weight loss showed that the effects of BCT combinations vary by behavioral focus (diet vs. physical activity).⁹⁰ While combining self-monitoring and goal-setting BCTs with other self-regulatory strategies have the potential to promote weight loss, the review concluded that there remains a lack of clarity regarding which packages of BCTs are most effective for weight loss and how BCTs might interact or dilute the effects of other BCTs.⁹⁰ Additional work should systematically evaluate combinations of BCTs that are effective for promoting multiple weight-management behaviors.

Providing feedback on the outcome of behaviors (e.g. weight change since last weighing occasion and statement about sticking to RF limit), when viewed, did not appear to be beneficial for promoting daily behavioral goal achievement. Given that frequent dietary self-monitoring and goal achievement are associated with improved weight management,^91–93 messages that directly cue participants to the association between their recent weight change and the impact of sticking to their goals could help them meet their goals more often, contributing to distal weight change. Previous research has demonstrated that feedback on outcome of behavior predicts positive long-term effects on diet and physical activity in adults with overweight or obesity,⁹⁴ however, less is known about effects on daily behavioral goal achievement. In the current study, we did not examine whether message effects varied in the context of more proximal weight change (i.e. gain or loss). Future work should examine moderators of message effects on daily outcomes.

Receipt of goal progress feedback messages was not associated with achievement of any goals compared to no receipt. This is similar to a recent microrandomized trial that tested the effect of goal progress messages on total daily steps and found no effect of message receipt compared to no message.³⁹ Given that most trackers and mobile programs provide progress indicators and goal reminders, it is unclear if this feedback is providing participants with new information. Messages in the current study were intended to draw attention to the discrepancy between current behavior and the daily goal, whether the current behavior was below or above the goal. While goal-setting and self-monitoring of behaviors are key behavior change strategies, there is limited evidence on whether frequent messages highlighting progress in relation to a goal are effective for promoting regular achievement of weight-related goals. In this trial, viewing goal progress feedback messages about RFs was associated with meeting RF goals suggesting that when participants view these messages, the knowledge of how close or far they are away from their RF goal may help them stay at or under it before the end of the day.

Being randomized to receive reminders to weigh, wear an activity tracker, and track RFs was not effective, though when viewed, reminder messages about RFs were associated with a higher likelihood of completing RF tracking. In general, self-monitoring was high (median of 72 weigh days out of 84, 80 tracker wear days, and 60 days with any RF tracking), and it is possible that participants did not need reminder messages to engage in these behaviors. It is also possible that since reminder messages were only delivered in the morning, participants may not have been receptive to early morning messages. Additionally, all the BCT-based messages may have been perceived to be prompts or cues to some degree since they all deployed app notifications.⁹⁵

The potential of JITAI messages and proximal goal achievement to improve distal weight loss was supported by findings that total number of message views and total number of behavioral goals achieved, both across and by individual behavior, were positively associated with 12-week weight loss. Further examination of ways to sustain message viewing, possible variation in effects over time and by individual contexts, are warranted. Overall weight loss after 12 weeks was modest at 3.0%, though similar to another trial in which participants used RF tracking and achieved 4.0% weight loss at 6 months.⁵⁶ Because this was a microrandomized trial with only one treatment group, and total message views is confounded by other aspects of treatment success even after use of SIPWs, it is not clear whether weight losses in this study were a result of the daily messages or other standard behavioral content that was available in the program (e.g. personalized weekly feedback on their progress). Given the clear importance of engagement in mobile programs, it will be necessary to study ways to increase engagement, including different types of messaging, interactive modules, and how to best deliver content and messaging (e.g. text versus video, long text pages vs. short modules, etc.). Both motivation to engage with intervention messages as well as the distinctive components of intervention fidelity (i.e. delivery, receipt, and enactment of messages) are important considerations that should be made when designing and evaluating future mHealth interventions.⁹⁶ Additional research is needed to elucidate the relationships among various measures of engagement beyond message views, such as user perceptions and psychological engagement, and intervention effects.⁹⁷

Strengths and limitations

A strength of this study was the microrandomized trial design which permitted frequent evaluation of message effects on multiple proximal weight-related behaviors, while prior studies have focused on single behavioral outcomes. The study design allowed for efficient testing of multiple BCT-based message types and examination of the effects of each of the seven message types on three different behaviors. Additionally, though this study had a small sample size, it was a diverse sample with 37% of young adults identifying with racial and ethnic minority backgrounds. Since individuals were excluded if they did not own iPhones, this may have resulted in a biased sample and limits generalizability of the results. Other limitations include the insufficient power to detect effects. For the most part, availability for each message type was lower than proposed in the power analyses, which likely affected the ability to answer the research questions. Expanding criteria for availability such that individuals are eligible to be randomized for each message type across all time points and regardless of their behavioral status (i.e. met goal or not) could improve power in future trials. Additionally, there were some technical issues with location tracking and system errors that lowered the overall number of messages that were available, particularly those that were specific to weighing. Careful testing of location tracking functions prior to study launch and ongoing monitoring of message fidelity and data integrity from study initiation could help identify technical issues earlier and reduce their impacts. While RF consumption was self-reported in the app and may have introduced systematic measurement error (e.g. under-reporting, recall bias), the criteria for meeting the RF limit required that participants had a complete day of tracking, which may have helped limit any systematic under-reporting. While each message focused on a distinct behavior, given the frequency of randomization in a single day, it is possible that any message received could have had carryover effects on other behaviors and prompted individuals to stay on track with achieving goals overall. With the exception of the evening messages, the current analysis focused on the effects of messages on goal achievement on the day of delivery and did not take into account the effects of messages that may have been received in prior days, which may have had lingering effects on goal achievement. The next step will be to conduct a trial with more participants and a longer duration and that allows all message types to be distributed at all time points; this will improve the ability to test the effect of specific BCT-based message types.

Conclusions and implications

Daily BCT-based messages may promote proximal goal achievement, but an important consideration is that only viewing messages (not being randomized to receive them) was associated with goal achievement, and this was likely confounded by other aspects of treatment success. This trial suggests that brief behavioral messages may only impact behavior if the participant reads them and calls attention to the role that participant motivation and contexts may play in participant receptivity to an intervention and its subsequent effects. Future research should consider the distinction between and importance of intervention delivery and receipt. As participant receptivity to and engagement with digital interventions often declines over time, examination of the time-varying effects of BCT-based messages on proximal achievement of weight-related goals is needed. Further optimization of JITAI messaging to support weight management may require collecting more information about participants’ context as potential moderators of message effects on outcomes (e.g. motivation, location, previous message dose, whether already met daily behavioral goal), varying message content or providing alternative incentives to increase engagement later in the program, and using other methods to determine how to send the “right message” at the “right time.”

Footnotes

Acknowledgements

The authors would like to thank the study participants for their contributions. We acknowledge the staff and students of the UNC Weight Research Program for their valuable support,including Julianne Powers,Sahil Vasa,Kayla Warechowski,Erin Coffman,and Lex Hurley,who provided research assistance.

ORCID iDs

Carmina G Valle

Brooke T Nezami

Nisha Gottfredson O’Shea

Karen E Hatley

Deborah F Tate

Ethical considerations

This study was approved by the Institutional Review Board (IRB) of The University of North Carolina at Chapel Hill (UNC) (IRB #16-0775) on 2 May 2018. All participants provided written informed consent prior to enrollment in the study.

Consent to participate

Informed consent was collected from participants prior to enrollment and data collection. Research staff guided participants through online consent forms which interested participants completed and submitted online and in a written paper copy.

Author contributions

CGV contributed to conceptualization,methodology,writing—original draft,writing—review & editing,supervision,and funding acquisition;BTN to conceptualization,methodology,data curation,writing—original draft,writing—review & editing,investigation,and project administration;NGO to formal analysis,writing—original draft,and writing—review and editing;KEH to methodology,supervision,and investigation;and DFT to conceptualization,methodology,writing—original draft,writing—review & editing,supervision,and funding acquisition.

Funding

The authors disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: This work was supported by the UNC Gillings School of Global Public Health (Gillings Innovation Lab award funded by the 2007 Gillings Gift);National Institute of Diabetes and Digestive and Kidney Disease (P30DK056350 to the UNC Nutrition Obesity Research Center to support UNC CHAI Core);National Cancer Institute (P30CA016086 to the UNC Lineberger Comprehensive Cancer Center to support UNC CHAI Core);National Center for Advancing Translational Sciences (NCATS) (UL1TR002489 to support REDCap at UNC). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Data availability statement

The datasets generated during and/or analyzed during the current study are not publicly available due to requirements to protect the privacy of participants,in accordance with their informed consent. However,de-identified data generated during and/or analyzed during the current study will be made available upon reasonable request to the corresponding author.

Guarantor

CGV.

Peer review

Dr. Lily Hawkins,University of Exeter.

References

Ward

Bleich

Cradockp

, et al. Projected U.S. state-level prevalence of adult obesity and severe obesity. N Engl J Med 2019; 381: 2440–2450.

Abdelaal

le Roux

Docherty

. Morbidity and mortality associated with obesity. Ann Transl Med 2017; 5: 161.

Lewis

Jacobs

McCreath

, et al. Weight gain continues in the 1990s: 10-year trends in weight and overweight from the CARDIA study. Coronary Artery Risk Development in Young Adults. Am J Epidemiol 2000; 151: 1172–1181.

Truesdale

Stevens

Lewis

, et al. Changes in risk factors for cardiovascular disease by baseline weight status in young adults who maintain or gain weight over 15 years: the CARDIA study. Int J Obes (Lond) 2006; 30: 1397–1407.

Hales

Carroll

Fryar

, et al. Prevalence of obesity and severe obesity among adults: United States, 2017–2018. NCHS Data Brief, no 360. Hyattsville, MD: National Center for Health Statistics., 27 February 2020.

Zheng

Manson

Yuan

, et al. Associations of weight gain from early to middle adulthood with major health outcomes later in life. JAMA 2017; 318: 255–269.

Keum

Greenwood

Lee

, et al. Adult weight gain and adiposity-related cancers: a dose-response meta-analysis of prospective observational studies. J Natl Cancer Inst 2015; 107: djv088.

Adams

Leitzmann

Ballard-Barbash

, et al. Body mass and weight change in adults in relation to mortality risk. Am J Epidemiol 2014; 179: 135–144.

Dietz

. Obesity and excessive weight gain in young adults: new targets for prevention. JAMA 2017; 318: 241–242.

10.

Jensen

Ryan

Apovian

, et al. 2013 AHA/ACC/TOS guideline for the management of overweight and obesity in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and The Obesity Society. Circulation 2014; 129: S102–S138.

11.

US Preventive Services Task Force, Curry SJ, Krist AH, et al. Behavioral weight loss interventions to prevent obesity-related morbidity and mortality in adults: US preventive services task force recommendation statement. JAMA 2018; 320: 1163–1171.

12.

Loria

Signore

Arteaga

. The need for targeted weight-control approaches in young women and men. Am J Prev Med 2010; 38: 233–235.

13.

Lytle

Svetkey

Patrick

, et al. The EARLY trials: a consortium of studies targeting weight control in young adults. Transl Behav Med 2014; 4: 304–313.

14.

Gokee-LaRose

Gorin

Raynor

, et al.

Are standard behavioral weight loss programs effective for young adults?

Int J Obes (Lond) 2009; 33: 1374–1380.

15.

LaRose

Tate

Lanoye

, et al. Adapting evidence-based behavioral weight loss programs for emerging adults: a pilot randomized controlled trial. J Health Psychol 2019; 24: 870–887.

16.

Pew Research Center. Demographics of mobile device ownership and adoption in the United States. Demographics of mobile device ownership and adoption in the United States, https://www.pewinternet.org/fact-sheet/mobile/ (2024, accessed March 4, 2024).

17.

LaRose

Guthrie

Lanoye

, et al. A mixed methods approach to improving recruitment and engagement of emerging adults in behavioural weight loss programs. Obes Sci Pract 2016; 2: 341–354.

18.

Wieland

Falzon

Sciamanna

, et al. Interactive computer-based interventions for weight loss or weight maintenance in overweight or obese people. Cochrane Database Syst Rev 2012: 8: CD007675.

19.

Riley

Rivera

Atienza

, et al.

Health behavior models in the age of mobile interventions: are our theories up to the task?

Transl Behav Med 2011; 1: 53–71.

20.

Laing

Mangione

Tseng

C-H

, et al. Effectiveness of a smartphone application for weight loss compared with usual care in overweight primary care patients: a randomized, controlled trial. Ann Intern Med 2014; 161: S5–S12.

21.

Hurkmans

Matthys

Bogaerts

, et al. Face-to-face versus mobile versus blended weight loss program: randomized clinical trial. JMIR Mhealth Uhealth 2018; 6: e14.

22.

Nahum-Shani

Smith

Spring

, et al. Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. Ann Behav Med 2018; 52: 446–462.

23.

Nahum-Shani

Hekler

Spruijt-Metz

. Building health behavior models to guide the development of just-in-time adaptive interventions: a pragmatic framework. Health Psychol 2015; 34S: 1209–1219.

24.

Nahum-Shani

Smith

Tewari

, et al. Just-in-time adaptive interventions (JITAIs): An organizing framework for ongoing health behavior support. University Park, PA: The Methodology Center, Penn State, 2014.

25.

Wang

Miller

. Just-in-the-moment adaptive interventions (JITAI): a meta-analytical review. Health Commun 2020; 35: 1531–1544.

26.

Klasnja

Smith

Seewald

, et al. Efficacy of contextually tailored suggestions for physical activity: a micro-randomized optimization trial of HeartSteps. Ann Behav Med 2019; 53: 573–582.

27.

Thomas

Bond

. Behavioral response to a just-in-time adaptive intervention (JITAI) to reduce sedentary behavior in obese adults: implications for JITAI optimization. Health Psychol 2015; 34S: 1261–1267.

28.

Hébert

Alexander

, et al. A mobile just-in-time adaptive intervention for smoking cessation: pilot randomized controlled trial. J Med Internet Res 2020; 22: e16907.

29.

Burke

Styn

Sereika

, et al. Using mHealth technology to enhance self-monitoring for weight loss: a randomized trial. Am J Prev Med 2012; 43: 20–26.

30.

Ambeba

Sereika

, et al. The use of mHealth to deliver tailored messages reduces reported energy and fat intake. J Cardiovasc Nurs 2015; 30: 35–43.

31.

Burke

Zheng

, et al. The SMARTER pilot study: testing feasibility of real-time feedback for dietary self-monitoring. Prev Med Rep 2017; 6: 278–285.

32.

Forman

Goldstein

Crochiere

, et al. Randomized controlled trial of OnTrack, a just-in-time adaptive intervention designed to enhance weight loss. Transl Behav Med 2019; 9: 989–1001.

33.

Dorsch

Cornellier

Poggi

, et al. Effects of a novel contextual just-in-time mobile app intervention (LowSalt4Life) on sodium intake in adults with hypertension: pilot randomized controlled trial. JMIR Mhealth Uhealth 2020; 8: e16696.

34.

Pfammatter

Marchese

Pellegrini

, et al. Using the preparation phase of the multiphase optimization strategy to develop a messaging component for weight loss: formative and pilot research. JMIR Formativ Res 2020; 4: e16297.

35.

Siopis

Chey

Allman-Farinelli

. A systematic review and meta-analysis of interventions for weight management using text messaging. J Hum Nutr Diet 2015; 28: 1–15.

36.

Klasnja

Hekler

Shiffman

, et al. Microrandomized trials: an experimental design for developing just-in-time adaptive interventions. Health Psychol 2015; 34S: 1220–1228.

37.

Klasnja

Rosenberg

Zhou

, et al. A quality-improvement optimization pilot of BariFit, a mobile health intervention to promote physical activity after bariatric surgery. Transl Behav Med 2021; 11: 530–539.

38.

Kramer

J-N

Künzler

Mishra

, et al. Which components of a smartphone walking app help users to reach personalized step goals? Results from an optimization trial. Ann Behav Med 2020; 54: 518–528.

39.

Figueroa

Deliu

Chakraborty

, et al. Daily motivational text messages to promote physical activity in university students: results from a microrandomized trial. Ann Behav Med 2022; 56: 212–218.

40.

Goldstein

Zhang

Klasnja

, et al. Optimizing a just-in-time adaptive intervention to improve dietary adherence in behavioral obesity treatment: protocol for a microrandomized trial. JMIR Res Protoc 2021; 10: e33568.

41.

NeCamp

Sen

Frank

, et al. Assessing real-time moderation for developing adaptive mobile health interventions for medical interns: micro-randomized trial. J Med Internet Res 2020; 22: e15033.

42.

Battalio

Conroy

Dempsey

, et al. Sense2Stop: a micro-randomized trial using wearable sensors to optimize a just-in-time-adaptive stress management intervention for smoking relapse prevention. Contemp Clin Trials 2021; 109: 106534.

43.

Yang

M-J

Sutton

Hernandez

, et al. A just-in-time adaptive intervention (JITAI) for smoking cessation: feasibility and acceptability findings. Addict Behav 2023; 136: 107467.

44.

Bidargaddi

Almirall

Murphy

, et al. To prompt or not to prompt? A microrandomized trial of time-varying push notifications to increase proximal engagement with a mobile health app. JMIR Mhealth Uhealth 2018; 6: e10123.

45.

Potter

N-SI

Lam

, et al. The mobile assistance for regulating smoking (MARS) micro-randomized trial design protocol. Contemp Clin Trials 2021; 110: 106513.

46.

Nordby

Gjestad

Kenter

RMF

, et al. The effect of SMS reminders on adherence in a self-guided internet-delivered intervention for adults with ADHD. Front Digit Health 2022; 4: 821031.

47.

Bell

Garnett

Qian

, et al. Notifications to improve engagement with an alcohol reduction app: protocol for a micro-randomized trial. JMIR Res Protoc 2020; 9: e18690.

48.

Nahum-Shani

Rabbi

Yap

, et al. Translating strategies for promoting engagement in mobile health: a proof-of-concept microrandomized trial. Health Psychol 2021; 40: 974–987.

49.

Michie

Ashford

Sniehotta

, et al. A refined taxonomy of behaviour change techniques to help people change their physical activity and healthy eating behaviours: the CALO-RE taxonomy. Psychol Health 2011; 26: 1479–1498.

50.

Michie

Richardson

Johnston

, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Ann Behav Med 2013; 46: 81–95.

51.

Tate

Lytle

Polzien

, et al. Deconstructing weight management interventions for young adults: looking inside the black box of the EARLY consortium trials. Obesity (Silver Spring) 2019; 27: 1085–1098.

52.

Michie

Abraham

Whittington

, et al. Effective techniques in healthy eating and physical activity interventions: a meta-regression. Health Psychol 2009; 28: 690–701.

53.

Olander

Fletcher

Williams

, et al. What are the most effective techniques in changing obese individuals’ physical activity self-efficacy and behaviour: a systematic review and meta-analysis. Int J Behav Nutr Phys Act 2013; 10: 29.

54.

Epstein

Paluch

Beecher

, et al. Increasing healthy eating vs. reducing high energy-dense foods to treat pediatric obesity. Obesity (Silver Spring) 2008; 16: 318–326.

55.

Nezami

Ward

Lytle

, et al. A mHealth randomized controlled trial to reduce sugar-sweetened beverage intake in preschool-aged children. Pediatr Obes 2018; 13: 668–676.

56.

Nezami

Hurley

Power

, et al. A pilot randomized trial of simplified versus standard calorie dietary self-monitoring in a mobile weight loss intervention. Obesity (Silver Spring) 2022; 30: 628–638.

57.

Tate

Lytle

Sherwood

, et al. Deconstructing interventions: approaches to studying behavior change techniques across obesity interventions. Transl Behav Med 2016; 6: 236–243.

58.

Shaffer

Diaz

Alcántara

, et al. An inexpensive device for monitoring patients’ weights via automated hovering. Int J Cardiol 2014; 172: e263–e264.

59.

Ross

Wing

. Concordance of in-home “smart” scale measurement with body weight measured in-person. Obes Sci Pract 2016; 2: 224–248.

60.

Evenson

Goto

Furberg

. Systematic review of the validity and reliability of consumer-wearable activity trackers. Int J Behav Nutr Phys Activity 2015; 12: 1–22.

61.

Leung

Case

Sung

M-C

, et al. A meta-analysis of Fitbit devices: same company, different models, different validity evidence. J Med Eng Technol 2022; 46: 102–115.

62.

Hardcastle

Jiménez-Castuera

Maxwell-Smith

, et al. Fitbit wear-time and patterns of activity in cancer survivors throughout a physical activity intervention and follow-up: exploratory analysis from a randomised controlled trial. PLoS ONE 2020; 15: e0240967.

63.

Wing

Tate

Espeland

, et al. Innovative self-regulation strategies to reduce weight gain in young adults: the study of novel approaches to weight gain prevention (SNAP) randomized clinical trial. JAMA Intern Med 2016; 176: 755–762.

64.

Wing

Tate

Espeland

, et al. Weight gain prevention in young adults: design of the study of novel approaches to weight gain prevention (SNAP) randomized controlled trial. BMC Public Health 2013; 13: 300.

65.

Liao

Klasnja

Tewari

, et al. Sample size calculations for micro-randomized trials in mHealth. Stat Med 2016; 35: 1944–1971.

66.

Luers

Klasnja

Murphy

. Standardized effect sizes for preventive mobile health interventions in micro-randomized trials. Prev Sci 2019; 20: 100–109.

67.

Seewald

Sun

Liao

. MRT-SS calculator: An R shiny application for sample size calculation in micro-randomized trials. arXiv.

68.

Pepe

Anderson

. A cautionary note on inference for marginal regression models with longitudinal data and general correlated response data. Commun Stat—Simul Comput 1994; 23: 939–951.

69.

Thissen

Steinberg

Kuang

. Quick and easy implementation of the Benjamini-Hochberg procedure for controlling the false positive rate in multiple comparisons. J Educ Behav Stat 2002; 27: 77–83.

70.

Hernán

Robins

(2020). Causal inference: what if. Boca Raton: Chapman & Hall/CRC. https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/ (accessed 31 May 2021).

71.

Valle

Nezami

Tate

. Designing in-app messages to nudge behavior change: lessons learned from a weight management app for young adults. Organ Behav Hum Decis Process 2020; 161: 95–101.

72.

Lyzwinski

. A systematic review and meta-analysis of mobile devices and weight loss with an intervention content analysis. J Pers Med 2014; 4: 311–385.

73.

Arigo

Brown

Pasko

, et al. Social comparison features in physical activity promotion apps: scoping meta-review. J Med Internet Res 2020; 22: e15642.

74.

Pelletier

Graham

Laska

. Social norms and dietary behaviors among young adults. Am J Health Behav 2014; 38: 144–152.

75.

Stok

de Vet

de Ridder

DTD

, et al. The potential of peer social norms to shape food intake in adolescents and young adults: a systematic review of effects and moderators. Health Psychol Rev 2016; 10: 326–340.

76.

Hawkins

Farrow

Thomas

. Do perceived norms of social media users’ eating habits and preferences predict our own food consumption and BMI? Appetite 2020; 149: 104611.

77.

Zeeni

Abi Kharma

Malli

, et al. Exposure to Instagram junk food content negatively impacts mood and cravings in young adults: a randomized controlled trial. Appetite 2024; 195: 107209.

78.

Arigo

Butryn

. Prospective relations between social comparison orientation and weight loss outcomes. Behav Med 2019; 45: 249–254.

79.

Carey

Connell

Johnston

, et al. Behavior change techniques and their mechanisms of action: a synthesis of links described in published intervention literature. Ann Behav Med 2019; 53: 693–707.

80.

Teixeira

Marques

Silva

, et al. A classification of motivation and behavior change techniques used in self-determination theory-based interventions in health contexts. Motiv Sci 2020; 6: 438–455.

81.

Sheeran

Listrom

Gollwitzer

. The when and how of planning: meta-analysis of the scope and components of implementation intentions in 642 tests. Eur Rev Soc Psychol 2024; 36: 162–194.

82.

Prestwich

Kellar

. How can the impact of implementation intentions as a behaviour change intervention be improved? Rev Eur Psychol Appl/Eur Rev Appl Psychol 2014; 64: 35–41.

83.

Hurley

Nezami

Valle

, et al. Motivated information avoidance in an mHealth weight loss intervention: associations between unmet behavioral goals and likelihood of viewing program messages. Digit Health 2024; 10: 1–9.

84.

Finne

Glausch

Exner

A-K

, et al. Behavior change techniques for increasing physical activity in cancer survivors: a systematic review and meta-analysis of randomized controlled trials. Cancer Manag Res 2018; 10: 5125–5143.

85.

Power

Tate

Valle

. Experiences of African American breast cancer survivors using digital scales and activity trackers in a weight gain prevention intervention: qualitative study. JMIR Mhealth Uhealth 2020; 8: e16059.

86.

Benyamini

Geron

Steinberg

, et al. A structured intentions and action-planning intervention improves weight loss outcomes in a group weight loss program. Am J Health Promot 2013; 28: 119–127.

87.

Hayes

Balantekin

Graham

, et al. Implementation intentions for weight loss in college students with overweight and obesity: a proof-of-concept randomized controlled trial. Transl Behav Med 2021; 11: 359–368.

88.

Luszczynska

Sobczyk

Abraham

. Planning to lose weight: randomized controlled trial of an implementation intention prompt to enhance weight reduction among overweight and obese women. Health Psychol 2007; 26: 507–512.

89.

Schroé

Van Dyck

De Paepe

, et al. Which behaviour change techniques are effective to promote physical activity and reduce sedentary behaviour in adults: a factorial randomized trial of an e- and m-health intervention. Int J Behav Nutr Phys Act 2020; 17: 127.

90.

Hawkins

Burns

Swancutt

, et al. Which components of behavioral weight management programs are essential for weight loss in people living with obesity? A rapid review of systematic reviews. Obes Rev 2024; 25: e13798.

91.

Payne

Turk

Kalarchian

, et al. Adherence to mobile-app-based dietary self-monitoring-impact on weight loss in adults. Obes Sci Pract 2022; 8: 279–288.

92.

Hutchesson

Tan

Morgan

, et al. Enhancement of self-monitoring in a web-based weight loss program by extra individualized feedback and reminders: randomized trial. J Med Internet Res 2016; 18: e82.

93.

MacLean

Wing

Davidson

, et al. NIH working group report: innovative research to improve maintenance of weight loss. Obesity (Silver Spring) 2015; 23: 7–15.

94.

Samdal

Eide

Barth

, et al. Effective behaviour change techniques for physical activity and healthy eating in overweight and obese adults; systematic review and meta-regression analyses. Int J Behav Nutr Phys Act 2017; 14: 42.

95.

Pirolli

Mohan

Venkatakrishnan

, et al. Implementation intention and reminder effects on behavior change in a mobile health system: a predictive cognitive model. J Med Internet Res 2017; 19: e397.

96.

Shaw

Steinberg

Zullig

, et al. Mhealth interventions for weight loss: a guide for achieving treatment fidelity. J Am Med Inform Assoc 2014; 21: 959–963.

97.

Short

DeSmet

Woods

, et al. Measuring engagement in eHealth and mHealth behavior change interventions: viewpoint of methodologies. J Med Internet Res 2018; 20: e292.

Outcomes of a pilot microrandomized trial of just-in-time adaptive intervention messages nudging weight-related behaviors in young adults

Abstract

Objective

Methods

Results

Conclusions

Trial registration

Keywords

Introduction

Methods

Study design and participants

Nudge intervention

Daily goals

Additional program features

Intervention messages based on behavior change techniques

Microrandomization of intervention messages

Measures

Proximal outcomes

Other measures

Statistical analyses

Results

Participants

Decision points and intervention messages delivered

Effects of receiving (vs. not receiving) any intervention messages

Effects of receiving (vs. not receiving) behavior-specific messages

Effects of receiving (vs. not receiving) BCT-based messages

Inverse probability weighted effects of viewing (vs. receiving and not viewing) any intervention messages

Inverse probability weighted effects of viewing (vs. receiving and not viewing) behavior-specific messages

Inverse probability weighted effects of viewing (vs. receiving and not viewing) BCT-based messages

Weight change and behavioral goal achievement

Associations between message views, goal attainment, and weight loss

Discussion

Strengths and limitations

Conclusions and implications

Footnotes

Acknowledgements

ORCID iDs

Ethical considerations

Consent to participate

Author contributions

Funding

Declaration of conflicting interests

Data availability statement

Guarantor

Peer review

References