Abstract
Objective
The SMARTER mobile health (mHealth) weight‐loss trial compared adherence to self‐monitoring (SM) of diet, physical activity (PA), and weight and adherence to study‐prescribed diet and PA goals between SM + feedback (SM + FB) and SM‐only arms over 12 months.
Methods
Participants used digital tools to monitor their dietary intake, PA, and weight. We applied generalized linear mixed modeling to compare patterns of monthly adherence to SM and behavioral goals between groups over time and examine the association of adherence to SM and behavioral goals with ≥5% weight loss.
Results
The sample (N = 502) was 80% female and 82% White, with a mean (SD) BMI of 33.7 (4.0) kg/m2. Adherence to SM and fat, calorie, and PA goals declined nonlinearly over time, with the SM + FB group displaying less of a decline compared with the SM‐only group. Higher adherence to diet, PA, and weight SM and to calorie and PA goals was associated with greater odds of achieving ≥5% weight loss. A higher monthly probability of achieving ≥5% weight loss was associated with greater adherence to diet, PA, and weight SM and to calorie and PA goals.
Conclusions
These results suggest that future research should examine the mechanisms underlying tailored FB to improve the effect of FB intervention strategies that can lead to improved weight loss.
Study Importance.
What is already known?
Providing proximal feedback (FB) to self‐monitoring (SM) can improve adherence, as well as adherence to behavior goals (diet, physical activity [PA]), and result in improved weight‐loss outcomes.
Use of digital tools for SM can reduce the burden of this behavior, particularly for PA and self‐weighing, whereas study participants report that recording food intake still requires an effort.
What does this study add?
Those who consistently engaged with the FB intervention had higher adherence to SM of diet, PA, and weight that was associated with improved weight loss; however, for those who did not monitor or open the messages regularly, the intervention dose was insufficient to promote sustained engagement with the digital tools used for SM.
The provision of remotely delivered FB for improving adherence may not have been effective for several reasons such as the timing of the message delivery if the participants were not near their phone, if they were not engaged with the use of the digital tools, or if message content was not relevant to them at that moment or time in the study.
How might these results change the direction of research or the focus of clinical practice?
Our findings demonstrate that remotely delivered FB alone is insufficient to sustain participants' engagement with the use of digital devices that facilitate SM and the improved weight‐loss outcomes that have been demonstrated to be associated with SM of diet, PA, and weight. The use of FB messages needs to be thoroughly examined in terms of content, frequency, timing, and mode of delivery.
Long‐term engagement with lifestyle‐focused digital tools remains a challenge that demands further examination of the most effective strategies to support use of the array of digital tools that are available and could benefit from input from interdisciplinary design teams that include laypersons.
INTRODUCTION
Obesity is a chronic disease associated with numerous comorbidities and affects ~42% of US adults [1]. Standard behavioral treatment (SBT) includes restricting energy intake, increasing energy expenditure, and employing behavioral strategies (e.g., goal setting, self‐monitoring [SM]) to develop healthful habits that support weight loss. Despite numerous improvements in SBT, weight‐loss results have remained mixed, especially when examining maintenance of weight loss [2, 3]. Mobile health (mHealth) approaches to SBT can reduce the burden of in‐person session attendance, and the use of digital tools can make SM easier [4, 5].
SM in weight loss is based on behavioral theory and extensive empirical evidence that SM is associated with superior weight‐loss outcomes [5, 6]. The ubiquity of smartphone apps for dietary recording, wearable devices for collection of physical activity (PA) data, and smart scales for self‐weighing provides opportunities for individuals to become engaged with SM as they develop new lifestyle habits. A systematic review of digital health SM found that SM is associated with superior weight loss [5]; however, evidence on engagement with these digital tools and adherence to SM strategies in behavioral weight‐loss studies has been mixed [7, 8]. Several of these studies have been short‐term and/or used small samples [7, 8, 9, 10].
In order to realize the full potential of mHealth interventions, we need to investigate how to improve adherence to SM when there is no in‐person component. In an earlier trial, we augmented SBT with a more limited feedback (FB) message system and observed enhanced adherence in the intervention group; however, the digital components were rudimentary [11]. Considering digital health advancements and the increasing prevalence of those who need weight‐loss treatment but may be unable or wish not to attend in‐person treatment, we designed a study that could leverage the technology to test a remotely delivered FB intervention in this subpopulation.
The primary aim of SMARTER, a two‐group randomized‐controlled weight‐loss trial, was to test the efficacy of a daily digital FB intervention delivered in response to SM data on dietary intake, PA, and weight (SM + FB) compared with SM only over 12 months. Findings showed that there was a statistically significant percentage of weight loss from baseline in both groups (−2.12% in SM + FB vs. −2.39% in SM only) but no difference between the groups [12].
We are reporting the findings of the trial's second aim, in which we hypothesized that real‐time FB would remind participants of the importance of SM and adhering to the behavioral change goals and, therefore, would improve adherence. The purpose of this investigation was as follows: 1) to compare adherence to SM of diet, PA, and weight and to dietary (i.e., fat and calorie) and PA goals between the treatment groups; 2) to assess the relationship between adherence to SM and study‐prescribed goals and weight loss; 3) to examine participant characteristics associated with digital tool engagement and adherence; and 4) to assess whether those who remained engaged and opened FB messages had better goal adherence and weight outcomes.
METHODS
SMARTER was approved by the institutional review board at the University of Pittsburgh and registered on ClinicalTrials.gov (NCT03367936). Inclusion criteria included body mass index (BMI) between 27 and 43 kg/m2, completion of a 5‐day electronic food diary, and ability to engage in moderate‐intensity PA. Exclusion criteria included needing medical supervision of diet or PA, pregnancy, serious mental illness, alcohol or eating disorder, and current weight‐loss treatment [4]. Recruitment for the study occurred from August 2018 to March 2020, and intervention delivery was completed in April 2021. Figure 1 illustrates the Consolidated Standards of Reporting Trials (CONSORT) diagram for SMARTER.
FIGURE 1.
CONSORT (Consolidated Standards of Reporting Trials) diagram. Previously published in Burke et al. [4]. FB, feedback; SM, self‐monitoring.
Intervention
All participants received one 90‐min, 1:1 in‐person session with a master's level prepared dietitian who had extensive experience with behavioral treatment of overweight and obesity. The dietitian also counseled participants using behavioral strategies to support weight reduction; goal setting (daily calorie and fat intake, PA minutes); daily SM of diet, PA, and weight; and the core concepts of SBT and demonstrated how to use the digital SM tools. For the SM + FB group, use of the SMARTER app and how to open the FB messages on participants' smartphones were demonstrated.
Dietary intake
Participants used the Fitbit app (Google LLC) to view food nutrient values, record food intake, and view daily intake summaries. The calorie goal was determined from the person's baseline body weight (women: 1200 kcal for <200 lb or 1500 kcal for ≥200 lb; men: 1500 kcal for <200 lb or 1800 kcal for ≥200 lb) and individualized as needed [4]. The fat gram goal approximated 25% of the calorie goal.
Physical activity
All participants monitored PA using a wrist‐worn Fitbit Charge 2 (Google) synced with their smartphone and interfaced with the study's informatics infrastructure [13]. Staff instructed participants to increase PA gradually, primarily by walking, and aim for 150 min/week by 12 weeks [4]. Once at goal, they were encouraged to add 10 min/week, with an ultimate goal of reaching 300 min/week by 52 weeks [4].
Weight
All participants were instructed to weigh daily on the study‐provided smart scale that transmitted weight data to their smartphone and the study database.
FB intervention
The provision of FB messages was limited to the SM + FB group. The SMARTER app was programmed to deliver up to three FB messages per day that were tailored to the available SM data and addressed caloric, fat, and added sugar intake daily and PA every other day. FB for self‐weighing occurred weekly. After opening the message, the participant could save it for future review; however, if participants did not open the FB message within 1 h after delivery, the message was no longer available. If participants were not syncing their smartphone with the Fitbit device, they may not receive FB messages.
Engagement with the digital SM tools was a crucial component; if the participant did not enter any SM data, messages were limited to encouraging the participant to engage in SM behaviors, as the algorithm had no SM data to address. In order to avoid participant desensitization to FB, we changed the message library at least monthly. FB to diet SM addressed calories, fat, and added sugar intake. Sample FB messages included:
“Calorie intake is above your goal, while fat grams are right on target. Take a moment to start to plan ahead for tomorrow.”
“A high fat intake can result in a high calorie intake. Focus on incorporating low fat foods and more fruits and vegetables the rest of the day.”
“Natural sugars are OK to eat, looks like you may have had some added sugars as well. Can you find any added sugar foods in your log?”
“Eating enough in the morning is important to fuel the rest of your day. Great effort recording your food intake today!”
Measures
Adherence to SM of diet, PA, and weight
Percentages of adherence to SM of diet, PA, and weight were calculated on a daily basis and averaged monthly. Adherence to diet SM was defined as recording ≥50% of daily calorie goals [4]. Adherence to PA SM was defined as recording ≥500 steps/day [14]. Adherence to weight SM was defined as having daily weight data obtained through the smart scale.
Adherence to fat, calorie, and PA goals
Percentages of adherence to fat and calorie goals were calculated on a daily basis and averaged monthly as follows:
and
Based on our previous weight‐loss research [15], we determined that having ≥15 days/month of diet SM data would provide the most representative estimation of an individual's daily food intake. We applied the same approach for the present analysis [16].
Weekly minutes of moderate‐ and vigorous‐intensity physical activities (MVPA) were calculated by summing respective Fitbit‐provided metabolic equivalents of fairly active (3–6 metabolic equivalents of task [MET]) and very active minutes (>6 MET), divided by the number of days with valid data (i.e., ≥4 days with ≥500 steps/day) and multiplied by 7 days.
Monthly percentages of adherence to PA goals were computed on a weekly basis and averaged by month as follows:
where, for the first 12 weeks, we prescribed PA goals of ≥150 min/week and gradually increased the prescribed PA goals to ≥300 min/week by 52 weeks. For percentage adherence to PA goals, values > 100% constitute adherence as exercising beyond the goals are acceptable.
Weight
Weight assessments were conducted by research staff at baseline and at 6 and 12 months using a research‐grade digital scale. Additionally, participants were provided a study‐issued smart scale to monitor their weight in their home setting and were advised to wear light clothing and no shoes when weighing. Monthly weight change percentages were calculated by comparing each month's average weight with baseline weight and expressed as a percentage. Participants were also categorized monthly into the following two groups based on weight loss: <5% versus ≥5% weight loss.
Intervention engagement
Monthly (or quarterly) percentages of FB messages opened were calculated as follows:
This intervention engagement variable was set to zero for the SM‐only group, as these participants did not receive tailored FB.
Participant characteristics associated with intervention engagement
Age, sex, and race were collected at baseline using a sociodemographic questionnaire. Baseline BMI (weight in kilograms divided by height in meters squared) was measured in person using the Tanita scale (Tanita Corp.).
Statistical analysis
We performed statistical analyses using SAS (version 9.4; SAS Institute Inc.). We described overall and monthly percentage of adherence to diet, PA, and weight SM and adherence to fat, calorie, and PA goals using means and standard deviations (SD) for the total sample and by treatment group.
We used linear mixed modeling to compare monthly adherence to diet, PA, and weight SM and adherence to fat, calorie, and PA goals over 12 months between treatment groups. Models included fixed effects for treatment, polynomial functions of time (linear, quadratic and cubic), and interactions of time with treatment, with random intercepts and slopes and an unstructured covariance matrix. Model assessment involved examining residuals, influence diagnostics, and model likelihood ratio test for overall fit. We omitted nonsignificant, higher‐order polynomial effects of time and their interactions with treatment to achieve model parsimony. Data transformations were applied to address non‐normally distributed residuals (reflected square root for SM of diet, PA, and weight, which reverses the interpretation to nonadherence, and square root for PA goals).
Similarly, for the relationship between each adherence variable and the monthly probability of achieving ≥5% weight loss, we used generalized linear mixed modeling with random intercepts, an unstructured covariance matrix, binary response distribution, and logit link. Models included fixed effects for the adherence variables for SM and behavioral goals as the primary independent variables and polynomial time effects (linear, quadratic, and cubic), as well as treatment, interactions of time with treatment, and interactions of adherence to SM or behavioral goals with treatment to account for study design effects. Nonsignificant, higher‐order effects were omitted for model parsimony.
In order to explore selected participant characteristics as predictors of receiving the FB intervention (opening ≥70% of FB messages) and meeting the SM adherence goals (≥70% of diet, PA, and weight SM) on a quarterly basis, we employed generalized linear mixed modeling with random intercepts, unstructured covariance matrix, binary response distribution, and logit link. Age, sex, race, education, income, and baseline BMI were included jointly as independent variables in the multivariable models.
RESULTS
The sample was mostly female (80%) and White (82%), with an average (SD) age of 45.0 (14.4) years and BMI of 33.7 (4.0) kg/m2. Participant characteristics were similar between treatment groups [15]. Over 12 months, the mean percentage of FB messages opened was 44.7 (28.8), and 22.9% of SM + FB participants met the FB protocol of opening ≥70% of FB messages. Overall, the mean percentages of days adherent to the SM goals were 51.6 (39.3) for diet SM, 78.2 (35.0) for PA SM, and 60.8 (34.9) for weight SM, with the percentages of participants achieving ≥70% adherence being 38.5%, 71.6%, and 47.5% for diet, PA, and weight SM, respectively. Additionally, mean percentages of days adherent to the behavioral goals were 21.5 (13.0), 45.8 (20.6), and 106.0 (82.1) for fat, calorie, and PA goals, respectively.
As described in Table 1 and shown in Figure 2 based on descriptive statistics, adherence to monthly diet, PA, and weight SM declined over 12 months, with similar curvilinear patterns by treatment group, whereas adherence to fat and calorie goals remained stable over time. Interestingly, the monthly percentage weight change declined in a similar pattern by treatment group over time. In general, both treatment groups achieved ~250 min/week of MVPA, with the SM + FB group showing slightly higher mean MVPA minutes per week and greater adherence to weekly PA goals compared with the SM‐only group but PA below the goal of ≤300 min/week from months 4 to 12 (Figure 3).
TABLE 1.
Monthly adherence to diet, PA, and weight SM and to fat, calorie, and PA goals for the total sample and by treatment group, as well as FB messages opened in the SM + FB group.
| Month | Total (N = 502) | SM (n = 251) | SM + FB (n = 251) | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Diet SM | PA SM | Weight SM | Fat goal | Calorie goal | PA goal | Diet SM | PA SM | Weight SM | Fat goal | Calorie goal | PA goal | Diet SM | PA SM | Weight SM | Fat goal | Calorie goal | PA goal | FB opened | |
| 1 | 81.7 (21.0) | 96.4 (8.7) | 82.7 (20.7) | 20.0 (12.2) | 49.0 (20.8) | 158.9 (101.5) | 78.5 (22.3) | 96.2 (8.6) | 82.8 (20.0) | 18.6 (11.2) | 47.8 (20.6) | 152.4 (99.4) | 85.0 (19.0) | 96.6 (8.8) | 82.7 (21.3) | 21.3 (12.9) | 50.2 (20.9) | 165.4 (103.4) | 71.6 (19.5) |
| 2 | 67.6 (32.2) | 91.6 (18.6) | 74.2 (27.3) | 21.7 (12.7) | 46.2 (20.0) | 156.8 (108.6) | 61.7 (33.8) | 91.4 (18.1) | 74.5 (27.3) | 20.4 (11.9) | 45.5 (20.6) | 151.2 (108.6) | 73.6 (29.4) | 91.8 (19.1) | 74.0 (27.4) | 22.8 (13.2) | 46.9 (19.5) | 162.5 (108.6) | 63.2 (26.1) |
| 3 | 60.7 (35.3) | 87.5 (24.3) | 68.7 (30.3) | 21.9 (12.2) | 46.4 (19.5) | 156.9 (107.2) | 53.4 (37.3) | 86.3 (24.9) | 68.8 (29.9) | 22.3 (11.0) | 46.7 (19.4) | 153.4 (104.6) | 67.9 (31.7) | 88.6 (23.8) | 68.6 (30.7) | 21.6 (13.0) | 46.1 (19.7) | 160.4 (109.8) | 56.9 (27.6) |
| 4 | 55.8 (36.6) | 83.3 (29.1) | 66.0 (31.5) | 22.2 (12.9) | 46.7 (20.0) | 82.5 (56.8) | 48.7 (38.0) | 81.9 (29.5) | 64.8 (32.0) | 22.4 (12.3) | 47.2 (21.6) | 79.4 (55.4) | 62.8 (33.8) | 84.6 (28.7) | 67.3 (31.1) | 22.0 (13.3) | 46.3 (18.9) | 85.7 (58.1) | 51.7 (28.4) |
| 5 | 51.0 (37.8) | 80.7 (32.0) | 63.0 (32.9) | 22.4 (13.7) | 44.3 (20.6) | 84.9 (59.8) | 45.4 (38.4) | 78.1 (32.9) | 62.2 (33.5) | 22.8 (12.5) | 44.1 (22.0) | 83.2 (59.0) | 56.6 (36.3) | 83.2 (30.8) | 63.8 (32.3) | 22.0 (14.7) | 44.5 (19.5) | 86.7 (60.7) | 47.0 (29.5) |
| 6 | 48.4 (39.1) | 78.0 (33.6) | 59.9 (34.4) | 21.7 (12.8) | 45.8 (20.7) | 85.5 (59.3) | 45.0 (39.7) | 77.5 (34.3) | 59.6 (35.1) | 21.6 (12.7) | 46.2 (22.8) | 83.9 (61.1) | 51.8 (38.3) | 78.4 (33.1) | 60.2 (33.6) | 21.9 (12.9) | 45.5 (18.9) | 87.0 (57.5) | 43.8 (30.4) |
| 7 | 48.0 (40.0) | 76.2 (36.3) | 58.5 (36.0) | 21.4 (13.3) | 45.9 (20.1) | 85.9 (57.8) | 45.9 (40.5) | 75.9 (35.8) | 59.1 (36.1) | 21.8 (12.6) | 48.0 (22.1) | 84.2 (57.8) | 50.1 (39.5) | 76.4 (36.9) | 57.9 (35.9) | 21.0 (13.9) | 44.0 (17.9) | 87.7 (58.0) | 39.8 (30.7) |
| 8 | 45.1 (40.1) | 73.6 (38.0) | 56.3 (35.8) | 21.3 (13.0) | 44.9 (20.7) | 87.5 (59.8) | 42.2 (40.0) | 74.3 (37.1) | 57.5 (36.0) | 21.0 (12.3) | 46.7 (23.2) | 84.8 (58.0) | 48.0 (40.1) | 73.0 (38.9) | 55.1 (35.6) | 21.6 (13.6) | 43.5 (18.2) | 90.3 (61.5) | 36.6 (30.9) |
| 9 | 42.8 (40.2) | 72.2 (40.0) | 53.1 (36.9) | 21.9 (14.3) | 44.0 (20.7) | 87.0 (59.6) | 40.0 (40.7) | 72.8 (39.9) | 53.9 (36.8) | 21.5 (14.6) | 42.8 (23.0) | 83.4 (57.3) | 45.5 (39.7) | 71.6 (40.2) | 52.3 (37.1) | 22.3 (14.0) | 45.0 (18.7) | 90.7 (61.8) | 33.6 (30.3) |
| 10 | 41.6 (40.2) | 69.5 (40.9) | 51.8 (36.8) | 21.6 (12.9) | 44.6 (21.8) | 86.3 (61.8) | 39.4 (40.8) | 69.7 (40.9) | 52.5 (37.3) | 21.3 (12.7) | 44.6 (24.0) | 83.9 (62.2) | 43.8 (39.6) | 69.3 (41.0) | 51.1 (36.3) | 21.9 (13.1) | 44.6 (19.8) | 88.8 (61.4) | 31.9 (30.5) |
| 11 | 40.0 (40.5) | 67.0 (41.8) | 49.0 (37.5) | 20.7 (13.8) | 42.6 (21.6) | 86.3 (58.7) | 37.5 (40.6) | 66.8 (41.6) | 49.6 (37.8) | 20.2 (13.4) | 42.7 (23.3) | 81.8 (56.4) | 42.4 (40.4) | 67.1 (42.1) | 48.4 (37.2) | 21.1 (14.2) | 42.6 (20.1) | 90.7 (60.7) | 31.0 (30.8) |
| 12 | 35.7 (39.4) | 62.2 (43.0) | 46.3 (37.0) | 21.9 (14.0) | 45.1 (21.6) | 84.6 (60.3) | 33.5 (39.0) | 62.8 (42.5) | 47.3 (37.3) | 20.4 (12.4) | 44.0 (23.6) | 79.8 (57.8) | 37.8 (39.7) | 61.7 (43.7) | 45.2 (36.8) | 23.3 (15.2) | 46.2 (19.6) | 89.4 (62.4) | 29.2 (30.6) |
| Total | 51.6 (39.3) | 78.2 (35.0) | 60.8 (34.9) | 21.5 (13.0) | 45.8 (20.6) | 106.0 (82.1) | 47.7 (39.6) | 77.9 (34.8) | 61.1 (35.0) | 21.0 (12.4) | 45.8 (21.9) | 102.5 (80.5) | 55.5 (38.6) | 78.5 (35.2) | 60.6 (34.9) | 21.9 (13.5) | 45.8 (19.5) | 109.5 (83.4) | 44.7 (28.8) |
Note: Total and monthly means and SD are reported for the total sample and by treatment group for the 12‐month period. Adherence to PA SM was computed as percentages of days when participants recorded ≥500 steps/day. Adherence to diet SM was computed as percentages of days when participants recorded ≥50% of their daily calorie goal. Adherence to weight SM was computed as percentages of days when participants recorded their weight daily. Abbreviations: FB, feedback; PA, physical activity; SM, self‐monitoring.
FIGURE 2.
Mean monthly percentage of SM of diet, PA, and weight and adherence to fat and calorie goals by treatment group over 12 months and percentage weight change from baseline. FB, feedback; PA, physical activity; SM, self‐monitoring. [Color figure can be viewed at wileyonlinelibrary.com]
FIGURE 3.
Mean monthly MVPA minutes and mean percentage adherence to PA goals by treatment group over 12 months. MVPA, moderate‐to‐vigorous intensity physical activity; PA, physical activity; SM, self‐monitoring; FB, feedback. [Color figure can be viewed at wileyonlinelibrary.com]
Percentages of days adherent to SM and behavioral goals by treatment group over 12 months
As displayed in Figure 4, the monthly percentages of days adherent to fat and calorie goals started low and changed nonlinearly over time, but these changes significantly differed between treatment groups (p < 0.05; Table S1). Compared with the SM‐only group, the SM + FB group started higher and had a less steep nonlinear change for adherence to fat and calorie goals.
FIGURE 4.
Linear mixed modeling predicted values of monthly percentage adherence to fat and calorie goals by treatment group over 12 months. SM, self‐monitoring; FB, feedback. [Color figure can be viewed at wileyonlinelibrary.com]
Figure 5 illustrates the nonlinear change in monthly percentages of days adherent to PA goals (square root‐transformed) and nonadherent to diet, PA, and weight SM (reflected square root‐transformed) by treatment group over 12 months. First, monthly percentages of days adherent to PA goals (square root‐transformed) declined nonlinearly over time (p < 0.05), with participants in the SM + FB group having a slightly higher adherence to PA goals compared with the SM‐only group over time (p < 0.05). Second, monthly percentages of days nonadherent to diet SM increased nonlinearly over time and differed significantly between treatment groups over 12 months, with the SM + FB group having lower levels of nonadherence compared with the SM‐only group for the first 6 months (p < 0.05; Table S1). Similarly, monthly percentages of days nonadherent to PA SM increased nonlinearly over time and varied by treatment group (p < 0.05). SM + FB participants had slightly lower levels of nonadherence to PA SM for the initial 4 months compared with the SM‐only group (p < 0.05). In contrast, monthly percentages of days nonadherent to weight SM increased nonlinearly over time, but this nonlinear increase did not differ between treatment groups (p ≥ 0.05; Table S1).
FIGURE 5.
Linear mixed modeling predicted values of monthly percentage adherence to diet, PA, and weight SM (reflected square root‐transformed) and PA goals (square root‐transformed) by treatment group over 12 months. FB, feedback; PA, physical activity; SM, self‐monitoring; SQRT, square root‐transformed. [Color figure can be viewed at wileyonlinelibrary.com]
The association between adherence to SM and behavioral goals and ≥5% weight loss
Table 2 summarizes the results of the association between monthly percentages of days adherent to SM and behavioral goals and achieving ≥5% weight loss monthly, considering each adherence variable separately. Higher adherence to diet, PA, and weight SM and to calorie goals was associated with greater odds of achieving ≥5% weight loss (Table 2). For adherence to PA goals, the relationship with achievement of ≥5% weight loss monthly was more complex, being dependent on treatment group assignment. Although greater adherence to PA goals tended to be associated with greater odds of achieving ≥5% weight loss, this association was weaker for the SM + FB group, as well as the probability, suggesting generally an increase followed by a curvilinear decrease over the 12‐month period. Although no significant treatment group differences were found for diet, PA, and weight SM, the achievement of ≥5% weight loss for fat and calorie goals varied between treatment groups over time, with the SM + FB group having reduced odds of achieving ≥5% weight loss compared with the SM‐only group.
TABLE 2.
Monthly odds of ≥5% weight loss with monthly percentages of SM and goal adherence over 12 months (N = 502)
| Model effect | Diet SM, AOR, 95% CI | PA SM, AOR, 95% CI | Weight SM, AOR, 95% CI | Fat goal, AOR, 95% CI | Calorie goal, AOR, 95% CI | PA goal, AOR, 95% CI |
|---|---|---|---|---|---|---|
| Adherence, % |
1.25, 1.19–1.31*** |
1.18, 1.11–1.24*** |
1.24, 1.18–1.31*** |
1.09, 0.97–1.24 |
1.10, 1.01–1.20* |
1.09, 1.05–1.12*** |
| Time, mo | ||||||
| Linear |
14.67, 9.83–21.89*** |
11.95, 8.07–17.70*** |
13.36, 8.98–19.89*** |
15.17, 9.58–24.02*** |
15.81, 9.97–25.06*** |
17.77, 11.15–26.46*** |
| Quadratic |
0.73, 0.68–0.78*** |
0.75, 0.70–0.80*** |
0.74, 0.69–0.79*** |
0.73, 0.67–0.78*** |
0.72, 0.67–0.78*** |
0.72, 0.67–0.77*** |
| Cubic |
1.012, 1.01–1.02*** |
1.011, 1.007–1.014*** |
1.011, 1.008–1.015*** |
1.012, 1.008–1.016*** |
1.012, 1.008–1.016*** |
1.012, 1.009–1.016*** |
| Treatment | ||||||
| SM + FB vs. SM (reference) |
0.80 0.45–1.43 |
0.93 0.52–1.67 |
0.98 0.55–1.74 |
1.53 0.71–3.28 |
1.55 0.72–3.32 |
2.60 1.01–6.69* |
| Treatment × Time | ||||||
| SM + FB × Time‐linear |
0.90 0.83–0.98* |
0.90 0.84–0.98* |
0.91 0.85–0.98* |
|||
| Adherence × Treatment | ||||||
| Adherence × SM + FB |
0.96 0.92–0.99* |
|||||
Note: Nonsignificant model effects were removed to achieve a more parsimonious generalized linear mixed model.
Abbreviations: AOR, adjusted odds ratio; FB, feedback; PA, physical activity; SM, self‐monitoring.
p < 0.05.
p < 0.001.
Patient characteristics as predictors of engagement
Table 3 summarizes the relationships between the odds of high engagement (i.e., achieving ≥70% adherence) in SM and age, sex, race, and baseline BMI on a quarterly basis over 12 months. Non‐White race was significantly related to lower odds of opening FB messages (p = 0.046). Male sex (p = 0.003), older age (p < 0.0001), and lower BMI (p < 0.0001) were significantly associated with higher odds of adherence to diet SM. Older age (p < 0.0001) and lower BMI were related to higher odds of adherence to weight SM (p = 0.008). Finally, male sex (p = 0.017), White race (p = 0.045), older age (p < 0.0001), and lower BMI (p = 0.047) were associated with greater odds of adherence to PA SM.
TABLE 3.
Quarterly odds of adherence to SM and opening FB messages with age, sex, race, and baseline BMI over 12 months (N = 502)
| Model effect | ≥70% FB messages opened, AOR, 95% CI | ≥70% diet, SM AOR, 95% CI | ≥70% weight, SM AOR, 95% CI | ≥70% PA SM, AOR, 95% CI |
|---|---|---|---|---|
| Time, quarterly | ||||
| 0–3 mo (reference) | ||||
| 4–6 mo |
0.33, 0.20–0.53*** |
0.23, 0.16–0.33*** |
0.30, 0.21–0.42*** |
0.60, 0.44–0.82*** |
| 7–9 mo |
0.16, 0.09–0.28*** |
0.16, 0.11–0.24*** |
0.17, 0.12–0.25*** |
0.58, 0.43–0.79*** |
| 10–12 mo |
0.11, 0.06–0.21*** |
0.10, 0.07–0.14*** |
0.09, 0.06–0.13*** |
0.30, 0.22–0.42*** |
| Age, 5‐y increment |
1.05, 0.99–1.05 |
1.40, 1.25–1.57** |
1.47, 1.28–1.69** |
1.28, 1.13–1.44** |
| Female vs. male (reference) |
0.88, 0.46–1.68 |
0.40, 0.22–0.73* |
0.70, 0.41–1.21 |
0.61, 0.40–0.92* |
| Non‐White vs. White (reference) |
0.39, 0.16–0.98* |
0.68, 0.35–1.34 |
0.70, 0.41–1.21 |
0.59, 0.35–0.99* |
| BMI, 5‐kg/m2 increment |
0.85, 0.70–1.04 |
0.68, 0.54–0.87** |
0.74, 0.58–0.94** |
0.82, 0.67–0.99* |
Note: Non‐White race category included African American, Asian American, Native Hawaiian, Pacific Islander, and multiracial individuals.
Abbreviations: AOR, adjusted odds ratio; FB, feedback; PA, physical activity; SM, self‐monitoring.
p < 0.05.
p < 0.01.
p < 0.001.
DISCUSSION
The study's aims were to examine long‐term engagement with digital tools, adherence to the core strategies of SBT, daily calorie and fat goals and weekly PA goals, and how adherence to SM and behavioral goals was associated with weight loss. Our primary analyses revealed that all SM adherence variables decreased nonlinearly over time. However, for fat and calorie goals, the SM + FB group showed higher adherence over time compared with the SM‐only group. Changes in adherence to diet and PA SM and to fat and calorie goals differed between treatment groups over time, with the SM + FB group showing a significant nonlinear decline; however, it was less so compared with the SM‐only group. In contrast, for adherence to weight SM, there were only significant time effects and no treatment main or interaction effects. Among those who were adherent to SM and behavioral goals, higher adherence to diet, PA, and weight SM and to calorie and PA goals was associated with greater odds of achieving ≥5% weight loss, except for fat goals. The probability of achieving ≥5% weight loss declined nonlinearly over time. Although no significant differences were found between treatment groups for adherence to diet, PA, and weight SM or to fat and calorie goals, the SM + FB group had higher average log odds of achieving ≥5% weight loss for PA goal adherence compared with the SM‐only group. Over time, the SM + FB group had smaller changes in the odds of ≥5% weight loss for fat, calorie, and PA goal adherence. A significant interaction was found between adherence to PA goals and treatment group on the odds of achieving ≥5% weight loss. Higher engagement with the FB protocol and greater adherence to SM and behavioral goals were associated with being older, male, and White and having a lower baseline BMI.
Several investigators have reported gradual declines in adherence behaviors; however, as observed in our trial as well as others, the decline varied across SM behaviors. Most recently, Carpenter et al. [8] reported disengagement from SM in a 6‐month trial testing two approaches to SM, with highest disengagement occurring for diet SM and self‐weighing. Even with digital tools, diet SM requires active engagement with greater effort and time to perform, unlike the more passive SM of wearing a PA tracker [5, 7]. Carpenter et al. found that most participants disengaged from diet SM, defined as a 2‐week period of no SM. More than 50% of those who had disengaged earlier reengaged in SM of weight or PA; however, less than 40% reengaged in diet SM. We did not examine periods of disengagement and reengagement in our 12‐month trial, but we did observe that PA tracking had the least decline over 12 months followed by self‐weighing; moreover, diet SM adherence had the most precipitous decline. These findings reflect the reduced effort required to perform some SM behaviors; however, using digital tools for dietary SM still requires time and effort.
There are significant challenges when comparing adherence with various SM behaviors or behavior‐change goals across weight‐loss studies, especially with the inconsistency in defining adherence to each behavior. We defined each measure of adherence based on empirical evidence from earlier trials [17, 18]. Examples of varying adherence definitions are dietary SM defined as “a day that any foods or beverages were logged,” PA SM defined as “a day that any steps were recorded,” and weight SM as “a day on which at least one weight value was captured” [8, 19]. Others have defined adherence as percent of days with recording [20, 21]. The significant variability in measuring adherence precludes comparing adherence outcomes across studies. Therefore, our reported adherence rates may be lower than comparative studies either because our definitions were more stringent and/or because our intervention study lasted 12 months, whereas several studies using digital tools lasted 6 months [5, 22]. Despite these differences in adherence measurement and study duration, what is consistent is that adherence declines over time and possibly declines further when the study period is longer.
Amagai et al. noted in a review of 69 mHealth studies using apps that there was no agreement on the definition of engagement; however, 95% of the studies measured some form of engagement around opening or using a specific app [23]. They identified the following three key app elements that supported engagement: tailored FB; personalized reminders; and app support from peers or coaches. Support from coaches is often cited as the most salient point for improved engagement in mHealth studies [24, 25, 26].
Other investigators have examined the effect of FB, with mixed results. West et al. [27] compared counselor‐crafted versus prescripted FB with group sessions (yes vs. no) and found that the group that received the briefer prescripted, modular FB had significantly better weight loss than the group that received the counselor‐crafted FB, with no difference in the intervention engagement. Most recently, Spring et al. [28] tested whether a wireless FB system (PA tracker, smart scale, and smartphone app to provide daily FB) resulted in less weight loss compared with a wireless FB system plus telephone coaching and a more intensive stepped‐up intervention if the participants did not respond to the initial treatment with greater weight loss. The weight changes at 3 and 12 months were comparable for the two groups, but a greater percentage of those in the coaching group achieved a 5% weight loss at 3 and 6 months that was not observable at 12 months.
The cumulative findings from these studies [27, 28], as well as our study, highlight the difference in response to FB that is delivered via phone versus a coach. The findings reveal the limitations of our understanding of the mechanisms underlying FB, i.e., increasing awareness of one's behavior or providing positive reinforcement for improved self‐regulation. Moreover, we need to examine the numerous elements of FB, including human versus app delivery, frequency, timing, message framing, extent of detail in FB, and how we can personalize FB while trying to scale the studies to reach larger and more diverse groups that need weight‐loss treatment.
Our study had many strengths, including a large sample size, randomized clinical trial design, high retention rate, use of validated measures, defined adherence metrics, objective measure of FB messages opened, and objective SM data. Our theory‐based intervention was expanded from a previously tested and efficacious FB system and used off‐the‐shelf digital tools. SMARTER is one of the largest trials of a digital weight‐loss intervention to report rates of adherence, goal attainment, and digital tool engagement to date. Limitations included enrollment of fewer male individuals and minorities than what was targeted, which limited generalizability, and lower‐than‐desired SMARTER app engagement.
CONCLUSION
The experimental intervention was the delivery of FB messages based on real‐time SM data; however, a small percentage of SM + FB participants opened the FB messages, resulting in receiving a low dose of the intervention. Our results suggest that the addition of FB to SM did not have a sustained impact on engagement with the SMARTER app and adherence to SM behaviors; however, those who remained engaged and opened more FB messages had better calorie goal adherence and weight outcomes.
Although the SMARTER app had some technical limitations affecting access to the message and receipt of FB, it is possible that the FB itself limited engagement and adherence to the study protocol. Declining adherence to SM resulted in messages to encourage SM instead of FB on progress made toward goals, which raises the issue of FB content. In addition to content, there are many nuances of a FB intervention that may have affected the SMARTER intervention, such as timing, personalization, and delivery methods. Future research needs to focus on the numerous aspects of FB messaging so that we can improve the efficacy of this critical intervention strategy.
CLINICAL TRIAL REGISTRATION
ClinicalTrials.gov identifier NCT03367936.
DISCLOSURE
The authors declared no conflicts of interest.
FUNDING INFORMATION
This research was supported by grants from the National Institutes of Health (NIH): R01HL131583 (LEB), R01HL1311583S (LEB, JKK), F31HL156278 (JC), and T32HL098048 (JC) and a grant from the University of Pittsburgh Clinical Translational Science Institute to Dr. S. Reis (UL1TR001857).
Supporting information
Table S1. Percentages of adherence to diet, PA and weight SM and study‐prescribed behavioral goals over 12 months.
Figure S1. Supporting Information.
ACKNOWLEDGEMENTS
The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the National Heart, Lung, and Blood Institute. Requests for additional information should be addressed to the Corresponding Author. Individual participant data collected during the trial, after de‐identification, will be available beginning 12 months and ending 36 months following article publication. We would like to thank the study participants for their time and contribution to the study, and the research staff who worked diligently to complete this trial in the context of the COVID‐19 restrictions.
Burke LE, Bizhanova Z, Conroy MB, et al. Adherence to self‐monitoring and behavioral goals is associated with improved weight loss in an mHealth randomized‐controlled trial. Obesity (Silver Spring). 2025;33(3):478‐489. doi: 10.1002/oby.24234
DATA AVAILABILITY STATEMENT
Data will be available within 6 months of publication date. Requests should be sent to the first author, L E Burke.
REFERENCES
- 1. Hales CM, Carroll MD, Fryar CD, Ogden CL. Prevalence of obesity and severe obesity among adults: United States, 2017‐2018. NCHS Data Brief. 2020;360:1‐8. [PubMed] [Google Scholar]
- 2. Butryn ML, Webb V, Wadden TA. Behavioral treatment of obesity. Psychiatr Clin North Am. 2011;34(4):841‐859. doi: 10.1016/j.psc.2011.08.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Elmaleh‐Sachs A, Schwartz JL, Bramante CT, Nicklas JM, Gudzune KA, Jay M. Obesity management in adults: a review. JAMA. 2023;330(20):2000‐2015. doi: 10.1001/jama.2023.19897 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Burke LE, Sereika SM, Parmanto B, et al. The SMARTER trial: design of a trial testing tailored mHealth feedback to impact self‐monitoring of diet, physical activity, and weight. Contemp Clin Trials. 2020;91:105958. doi: 10.1016/j.cct.2020.105958 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Patel ML, Wakayama LN, Bennett GG. Self‐monitoring via digital health in weight loss interventions: a systematic review among adults with overweight or obesity. Obesity (Silver Spring). 2021;29(3):478‐499. doi: 10.1002/oby.23088 [DOI] [PubMed] [Google Scholar]
- 6. Burke LE, Wang J, Sevick MA. Self‐monitoring in weight loss: a systematic review of the literature. J Am Diet Assoc. 2011;111(1):92‐102. doi: 10.1016/j.jada.2010.10.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Butryn ML, Martinelli MK, Crane NT, et al. Counselor surveillance of digital self‐monitoring data: a pilot randomized controlled trial. Obesity (Silver Spring). 2020;28(12):2339‐2346. doi: 10.1002/oby.23015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Carpenter CA, Eastman A, Ross KM. Consistency with and disengagement from self‐monitoring of weight, dietary intake, and physical activity in a technology‐based weight loss program: exploratory study. JMIR Form Res. 2022;6(2):e33603. doi: 10.2196/33603 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Szinay D, Forbes CC, Busse H, DeSmet A, Smit ES, Konig LM. Is the uptake, engagement, and effectiveness of exclusively mobile interventions for the promotion of weight‐related behaviors equal for all? A systematic review. Obes Rev. 2023;24(3):e13542. doi: 10.1111/obr.13542 [DOI] [PubMed] [Google Scholar]
- 10. Varela C, Oda‐Montecinos C, Andres A, Saldana C. Effectiveness of web‐based feedback interventions for people with overweight and obesity: systematic review and network meta‐analysis of randomized controlled trials. J Eat Disord. 2021;9(1):75. doi: 10.1186/s40337-021-00432-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Burke LE, Styn MA, Glanz K, et al. SMART trial: a randomized clinical trial of self‐monitoring in behavioral weight management‐design and baseline findings. Contemp Clin Trials. 2009;30(6):540‐551. doi: 10.1016/j.cct.2009.07.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Burke LE, Sereika SM, Bizhanova Z, et al. The effect of tailored, daily, smartphone feedback to lifestyle self‐monitoring on weight loss at 12 months: the SMARTER randomized clinical trial. J Med Internet Res. 2022;24(7):e38243. doi: 10.2196/38243 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Pulantara IW, Wang Y, Burke L, et al. Streamlining data collection and management of mHealth, wearable, and IoT in digital behavioral health intervention with ADAM: a novel informatics architecture (preprint). JMIR Mhealth Uhealth. 2023;12:1–13. doi: 10.2196/50043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. St Fleur RG, St George SM, Leite R, Kobayashi M, Agosto Y, Jake‐Schoffman DE. Use of Fitbit devices in physical activity intervention studies across the life course: narrative review. JMIR Mhealth Uhealth. 2021;9(5):e23411. doi: 10.2196/23411 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Burke LE, Sereika SM, Parmanto B, et al. Effect of tailored, daily feedback with lifestyle self‐monitoring on weight loss: the SMARTER randomized clinical trial. Obesity (Silver Spring). 2022;30(1):75‐84. doi: 10.1002/oby.23321 [DOI] [PubMed] [Google Scholar]
- 16. Stote KS, Radecki SV, Moshfegh AJ, Ingwersen LA, Baer DJ. The number of 24 h dietary recalls using the US Department of Agriculture's automated multiple‐pass method required to estimate nutrient intake in overweight and obese adults. Public Health Nutr. 2011;14(10):1736‐1742. doi: 10.1017/S1368980011000358 [DOI] [PubMed] [Google Scholar]
- 17. Acharya SD, Elci OU, Sereika SM, Styn MA, Burke LE. Using a personal digital assistant for self‐monitoring influences diet quality in comparison to a standard paper record among overweight/obese adults. J Am Diet Assoc. 2011;111(4):583‐588. doi: 10.1016/j.jada.2011.01.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Burke LE, Styn MA, Sereika SM, et al. Using mHealth technology to enhance self‐monitoring for weight loss: a randomized trial. Am J Prev Med. 2012;43(1):20‐26. doi: 10.1016/j.amepre.2012.03.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Painter SL, Ahmed R, Hill JO, et al. What matters in weight loss? An in‐depth analysis of self‐monitoring. J Med Internet Res. 2017;19(5):e160. doi: 10.2196/jmir.7457 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Goldstein SP, Goldstein CM, Bond DS, Raynor HA, Wing RR, Thomas JG. Associations between self‐monitoring and weight change in behavioral weight loss interventions. Health Psychol. 2019;38(12):1128‐1136. doi: 10.1037/hea0000800 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Jospe MR, Roy M, Brown RC, et al. The effect of different types of monitoring strategies on weight loss: a randomized controlled trial. Obesity (Silver Spring). 2017;25(9):1490‐1498. doi: 10.1002/oby.21898 [DOI] [PubMed] [Google Scholar]
- 22. Turner‐McGrievy GM, Wilcox S, Boutté A, et al. The dietary intervention to enhance tracking with Mobile devices (DIET Mobile) study: a 6‐month randomized weight loss trial. Obesity (Silver Spring). 2017;25(8):1336‐1342. doi: 10.1002/oby.21889 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Amagai S, Pila S, Kaat AJ, Nowinski CJ, Gershon RC. Challenges in participant engagement and retention using Mobile health apps: literature review. J Med Internet Res. 2022;24(4):e35120. doi: 10.2196/35120 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Spring B, Pellegrini CA, Pfammatter A, et al. Effects of an abbreviated obesity intervention supported by mobile technology: the ENGAGED randomized clinical trial. Obesity (Silver Spring). 2017;25(7):1191‐1198. doi: 10.1002/oby.21842 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Thomas JG, Bond DS, Raynor HA, Papandonatos GD, Wing RR. Comparison of smartphone‐based behavioral obesity treatment with gold standard group treatment and control: a randomized trial. Obesity (Silver Spring). 2019;27(4):572‐580. doi: 10.1002/oby.22410 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Antoun J, Itani H, Alarab N, Elsehmawy A. The effectiveness of combining nonmobile interventions with the use of smartphone apps with various features for weight loss: systematic review and meta‐analysis. JMIR Mhealth Uhealth. 2022;10(4):e35479. doi: 10.2196/35479 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. West DS, Krukowski RA, Stansbury ML, Ogden D, Borden J, Harvey JR. Examining weekly facilitated group sessions and counselor‐crafted self‐monitoring feedback on treatment outcome in digital weight control: a pilot factorial study. Obes Sci Pract. 2022;8(4):433‐441. doi: 10.1002/osp4.585 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Spring B, Pfammatter AF, Scanlan L, et al. An adaptive behavioral intervention for weight loss management: a randomized clinical trial. JAMA. 2024;332(1):21‐30. doi: 10.1001/jama.2024.0821 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Percentages of adherence to diet, PA and weight SM and study‐prescribed behavioral goals over 12 months.
Figure S1. Supporting Information.
Data Availability Statement
Data will be available within 6 months of publication date. Requests should be sent to the first author, L E Burke.