ABSTRACT
Objective
Behavioral weight management programs (BWLPs) encourage participants to self‐monitor dietary intake daily, preferably throughout the day as eating/drinking occurs. Despite empirical support for daily self‐monitoring, little is known regarding within‐day patterns of self‐monitoring and their associations with weight loss.
Methods
Within‐day patterns of self‐monitoring and associations with weight loss were examined in 449 adults with obesity (mean ± SD age = 49.4 ± 11.4 years, BMI = 35.7 ± 4.0 kg/m2, 83.5% female, 23.4% Black or African American, 9.8% Hispanic) who enrolled in a 16 week BWLP and were asked to self‐monitor dietary intake daily using a smartphone app.
Results
K‐means cluster analysis resulted in four patterns: participants who primarily engaged in Minimal Logging (n = 101), Backlogging (n = 124), Logging one occasion/day (n = 59), and Logging two or more occasions/day (n = 165). There was not a difference in weight loss between Logging two or more occasions/day (M ± SD; 9.1 ± 4.1%) and Logging one occasion/day (7.7 ± 4.7%), p = 0.219; however, both lost more weight than Backlogging (5.7 ± 4.1%) and Minimal logging (2.1 ± 3.4%), ps < 0.001. Moreover, 88.5% of Logging two or more occasions/day participants achieved clinically‐meaningful weight losses, compared to 71.2% of Logging one occasion/day, 55.6% of Backlogging, and 19.8% of Minimal Logging, all ps < 0.05.
Conclusions
Results support recommendations for daily (and, ideally, throughout‐the‐day) self‐monitoring of dietary intake. Future studies should investigate strategies to increase self‐monitoring adherence, including technological supports and lower‐burden tools.
Keywords: dietary intake, obesity, overweight, self‐monitoring, weight change
1. Introduction
Self‐monitoring of dietary intake is a core component in existing behavioral weight‐loss programs (BWLPs) [1, 2]. In these programs, participants are asked to track all foods and drinks consumed each day, either using paper logs [2] or newer technologies such as smartphone food diary apps [2, 3]. Per self‐regulation theory [4], self‐monitoring allows participants to observe their dietary behaviors and track progress toward program goals (e.g., reducing caloric intake); observation of goals being met provides reinforcement and observation of discrepancies allows individuals to identify areas for change (e.g., noting that a specific type of meal or pattern of snacking makes it harder to meet a one's calorie goal). Importantly, the empirical literature also supports the importance of self‐monitoring, as greater adherence to self‐monitoring of dietary intake has been associated with greater weight loss (both across the course of interventions [2, 5, 6, 7, 8] and more proximally, on a within‐week level [9]), mediated by greater calorie goal attainment [10].
Beyond the instruction to self‐monitor dietary intake daily, BWLP participants are also often encouraged to self‐monitor throughout the day, as soon as possible after consumption of foods/drinks [8]. Theoretically, this should help participants improve calorie goal attainment by allowing modifications of food/drink choices throughout the day in response to self‐monitoring data (e.g., using a calorie goal as a declining balance and changing food choices for dinner after a larger‐than‐expected lunch). Moreover, logging throughout the day increases the accuracy of self‐monitoring data by decreasing risk of recall bias [11]. As a related strategy, BWLPs also often introduce the strategy of “pre‐planning,” or looking up the nutritional information for foods/drinks before they are consumed to help plan out meals/future food and drink choices; this could help individuals engage in problem solving, developing plans to deal with challenging situations (e.g., deciding in advance what to eat during a work party vs. logging all foods/drinks consumed after the party, when the opportunity to make changes has passed). Days of “pre‐logging,” logging foods/drinks on days before they are consumed, may capture (in part) this pre‐planning process.
Despite how common these strategies are within BWLPs, there has been little empirical evidence to support their use. In one of the first studies that tried to investigate the associations between timing of self‐monitoring and weight loss outcomes, Burke et al. [6] asked 35 BWLP participants to self‐monitor dietary intake using instrumented paper diaries that recorded when they were opened. Participants who had a higher proportion of entries recorded within 15 min of the reported consumption time experienced greater weight loss at 6 months. More recently, Tang et al. [12] also found that greater timeliness of self‐monitoring using a study‐provided app (scored as completing self‐monitoring records not only for dietary intake but also physical activity, sedentary behavior, mood, and weight loss satisfaction at either the requested time, later on the same day, the next day, or not at all) was associated with greater weight loss in 61 adults enrolled in a 5‐week, online BWLP. Similarly, using data from an Internet‐based weight loss program, Harvey et al. [8] found that participants who lost at least 5% of their initial body weight logged into the study website to self‐monitor caloric intake more often (an average of 2.4 times/day) than those who lost < 5% (1.6 times/day). Finally, Hayes et al. [13] used a questionnaire querying meal planning behavior and found that participants in a worksite weight loss program that reported more frequent meal planning lost more weight; to date, however, no studies have investigated the behavior of “pre‐logging” foods/drinks in self‐monitoring logs prior to their consumption.
The emergence of smartphone applications for self‐monitoring of dietary intake has allowed for a level of rich data capture (e.g., providing time‐stamped data entries, synced directly with cloud servers) that allow analyses at levels not previously possible. The current study used a rich longitudinal dataset, collected from 449 participants who were asked to use a smartphone self‐monitoring app to track their dietary intake throughout a 16 week BWLP, to address the gaps in the empirical literature related to within‐day patterns of dietary self‐monitoring behavior. The primary aim was to identify patterns in dietary self‐monitoring; for this, data were coded for timeliness of self‐monitoring (i.e., as pre‐logging, logging on two or more occasions/day, logging on one occasion/day, backlogging, or not logging) and K‐means cluster analysis was used to identify discrete clusters of dietary self‐monitoring patterns over the 16 week BLWP. The second aim of this study was to identify whether these patterns of self‐monitoring were differentially associated with weight loss during the 16 week BWLP. It was hypothesized that participants who more often self‐monitored their dietary intake prior to or throughout the day of record would demonstrate significantly greater percent weight loss at the end of the intervention compared to those who self‐monitored once a day, on a later day, or not at all. As exploratory aims, differences between clusters in (1) the proportion of participants who experienced clinically‐meaningful weight losses (≥ 5% reduction from baseline) and (2) demographics and baseline characteristics (i.e., age, BMI, gender, race, ethnicity, marital status, level of education and annual household income) were explored.
2. Methods
The current study analyzed intervention data from the Support and Tracking to Achieve Results (Project STAR) clinical trial (NCT04116853), which aimed to evaluate the impact of an adaptive intervention on long‐term weight loss maintenance [14]. Prior to trial randomization, Project STAR participants received a 16 week BWLP; only those who lost ≥ 5% of their baseline weight at the post‐intervention assessment were eligible for randomization into one of the two maintenance programs. The current study used data only from the initial BWLP, prior to final eligibility assessment and randomization. The BWLP was conducted across 5 cohorts, with start dates staggered between January 2020 and October 2022. All study procedures were approved by the University of Florida Institutional Review Board.
2.1. Participants
Full recruitment methods (including inclusion/exclusion criteria), details regarding participant demographics and baseline characteristics, and details regarding initial BWLP outcomes have been published previously [15]. In brief, the participants were 449 adults with obesity (mean ± SD age = 49.5 ± 11.4 years, BMI = 35.7 ± 4.0 kg/m2) who owned a smartphone and reported no contraindications to participation in a group‐based BWLP. Majority (n = 375, 83.5%) of participants identified as female and, in terms of race and ethnicity, 74.2% (n = 333) identified as White, 23.4% (n = 105) as Black or African American, 9.8% (n = 44) as Hispanic, 3.1% (n = 14) as Asian, 1.6% (n = 7) as American Indian or Alaskan Native, and 0.4% (n = 2) as Native Hawaiian or Other Pacific Islander (participants could select more than one race/ethnicity, thus totals exceed 100%).
2.2. BWLP Intervention
Participants were provided with a 16‐week group‐based BWLP modeled after the lifestyle intervention from the Diabetes Prevention Program [16]. At the first session, participants were provided with initial calorie goals (1200–1800 kcal/day depending on baseline weight), taught how to use study‐provided tools (a BodyTrace e‐scale and the FatSecret smartphone app) to self‐monitor their weight and dietary intake, and encouraged to self‐monitor both weight and dietary intake each day throughout the intervention period. Participants were encouraged to track all/foods drinks consumed throughout the day as close as possible to the time at which consumption occurred; this was explained as a way to improve accuracy of self‐monitoring records and thus how helpful they would be for guiding changes in dietary habits. In the fourth session, pre‐planning (and pre‐logging) meals was discussed as one of several strategies for helping participants manage challenging or “high risk” situations (i.e., situations in which they may have difficulty meeting their calorie goal, such as at family events or when eating out at restaurants).
At the fifth session, participants were provided with physical activity goals (gradually increasing engagement in moderate‐intensity physical activity, such as brisk walking, up to 150–300 min/week) and asked to start self‐monitoring their physical activity each day via the FatSecret smartphone application. Assuming that participants who did not return at Month 4 (n = 31, 6.9% of those enrolled) regained ∼0.3 kg/month (calculated as 0.01 kg/day) from their last measured e‐scale weight (an assumption used in several prior BWLP trials [17, 18]), participants lost an average (mean ± SD) of 6.4 ± 4.9% of their baseline weight at Month 4.
2.3. Measures
Demographic characteristics were assessed via a self‐report questionnaire at baseline. Weight was assessed at baseline and the post‐intervention assessment via study‐provided e‐scales (BodyTrace Inc.) that sent weights to research servers via the cellular network. These e‐scales have demonstrated good concordance with assessment weights measured in‐person [19, 20] and a standardized assessment protocol was used [21]. Dietary self‐monitoring data were collected via the FatSecret Platform API [22]; data were pulled every 15 min, and a timestamp was automatically applied by the backend data capture system. Each participant had an entry for each item entered, coded with the date and time that the data were pulled, the date that the food/drink was reported as consumed, the meal each food/drink was categorized under (e.g., breakfast, lunch, dinner, snack/other) and nutrition information (e.g., portion size, macronutrients, calories). If an item that was previously imported was adjusted by the participant at a later time or date, the timestamp was overwritten. For example, if a participant returned to a prior day's dietary record and changed a food item that had been previously entered, then that original item's timestamp was replaced with the most recent entry's timestamp.
2.4. Data Analyses
A custom python script was used to code each intervention day for each participant within one of the following five categories: (1) ≥ 1 item self‐monitored prior to the day of record, no items monitored on the day of record or any later dates (i.e., pre‐logging); (2) all items self‐monitored on the day of record with ≥ 2 items recorded > 1 h apart (i.e., logging on two or more occasions/day); (3) all items self‐monitored on the day of record with all self‐monitoring occurring within an 1 h range (i.e., logging on one occasion/day); (4) ≥ 1 item logged on a day after the day of record (i.e., backlogging); (5) no items self‐monitored (i.e., no logging). Each day was coded into only one category; for example, a participant may have logged in Monday morning to self‐monitor all of the foods/drinks they consumed on Sunday, and then later that evening logged in again to record all of the foods/drinks they ate over the course of that same day (i.e., on that Monday). The Sunday entry would be categorized as backlogging (i.e., ≥ 1 item logged on a day after the day of record) and Monday as logging on one occasion/day (i.e., self‐monitoring occurred within a 1 h range on the day of record).
Statistical analyses were conducted using SPSS version 29.0 and R version 4.5.0. K‐means cluster analysis was used to identify subgroups of participants according to the proportion of days that participants engaged in each of the 5 categories of self‐monitoring described previously. Final model selection balanced parsimony with complexity, and a final model was selected with respect to iteration history, intercluster distance (higher is better), intracluster distance (lower is better), Dunn's Index, silhouette scores, and visual appraisal of dendrograms. An ANOVA was used to assess differences in percent weight change by cluster, with Games‐Howell post hoc tests used to identify differences between clusters due to unequal variances. A chi‐square test was used to examine the difference in the proportion of participants in each cluster achieving weight losses of ≥ 5% from baseline. Additional ANOVAs (for continuous variables) and chi‐square tests (for categorical variables, with Fisher's exact test used for analyses with cell counts < 5) were used to examine differences in demographic and baseline characteristics between clusters.
3. Results
Of the 49,839 possible days of dietary self‐monitoring across the overall sample (111 out of 112 intervention days, as participants were only asked to start self‐monitoring the day after their first intervention session), there was at least one item logged on 39,485 days (79.2% of possible days). Participants did not log at all on 10,354 days (20.8%), backlogged on 14,546 days (29.2%), logged on one occasion/day on 8802 days (17.7%), logged on two or more occasions/day on 15,221 days (30.5%), and pre‐logged on 916 days (1.8%). Figure 1 demonstrates the proportion of participant days in each category of self‐monitoring over each month of the intervention; adherence to self‐monitoring was highest in the first month of the program and decreased over time. Table 1 provides correlations between the number of days that participants engaged in each type of logging and weight change during the BWLP.
FIGURE 1.
Proportion of participant days coded in each self‐monitoring category during each month of the 16 week behavioral weight loss program.
TABLE 1.
Correlations between the number of days coded in each dietary self‐monitoring category and percent weight loss during the 16 week behavioral weight loss program.
| Variable | Days did not log | Days backlogged | Days logged one occasion/day | Days logged two or more occasions/day | Days pre‐logged | Percent weight change from baseline to month 4 |
|---|---|---|---|---|---|---|
| Days did not log | — | |||||
| Days backlogged | −0.33*** | — | ||||
| Days logged one occasion/day | −0.26*** | −0.06 | — | |||
| Days logged two or more occasions/day | −0.53*** | −0.35*** | −0.37*** | — | ||
| Days pre‐logged | −0.17*** | −0.011* | −0.11* | 0.16*** | — | |
| Percent weight change from baseline to month 4 | 0.57*** | 0.02 | −0.14** | −0.46*** | −0.08 | — |
p < 0.05.
p < 0.01.
p < 0.001.
Based on fit indices (see Table 2) and parsimony, a 4‐cluster model was selected. Figure 2 displays the proportion of participant days coded into each self‐monitoring category by cluster. Participants in Cluster 1 (n = 101; 22.4%) were characterized mostly by no logging and thus this cluster will henceforth be referred to as the “Minimal Logging” group. Cluster 2 (n = 124; 27.6%) was mostly characterized by “Backlogging,” Cluster 3 (n = 59; 13.1%) was mostly characterized by “Logging one occasion/day,” and Cluster 4 (n = 165; 36.7%) was mostly characterized by “Logging two or more occasions/day.”
TABLE 2.
Fit indices for K‐means cluster analysis models modeling 2–5 potential clusters.
| Indices of fit | 2 clusters | 3 clusters | 4 clusters | 5 clusters |
|---|---|---|---|---|
| Iteration history | 9 | 5 | 3 a | 5 |
| Intercluster distance | 77.70 | 78.93 | 79.13 a | 76.19 |
| Intracluster distance | 17,672 | 13,026 | 10,936 | 9620 a |
| Dunn's index | 0.004 | 0.018 | 0.043 | 0.079 a |
| Silhouettes | Good | Good | Good | Good |
| Participants per cluster | 345 | 165 | 101 | 103 |
| 104 | 101 | 124 | 95 | |
| 183 | 59 | 55 | ||
| 165 | 97 | |||
| 99 |
The most ideal fitting parameter.
FIGURE 2.
Proportion of participant days coded in each self‐monitoring category by cluster.
There was a significant difference in percent weight change during the BWLP by cluster, F(3, 288.8) = 63.04, p < 0.001 (Brown‐Forsythe F‐ratio reported due to a significant Levene's test). Specifically, there was not a significant difference in weight loss between the Logging two or more occasions/day (M ± SD; 9.1 ± 4.1%) and Logging one occasion/day (7.7 ± 4.7%) groups (see Table 3); however, participants in both the Logging two or more occasions/day and Logging one occasion/day groups lost more weight than those in the Backlogging (5.7 ± 4.1%) and Minimal Logging (2.1 ± 3.4%) groups. There was also a significant difference between clusters in terms of the proportion of participants achieving clinically‐meaningful weight losses, X 2(3) = 129.28, p < 0.001, with 88.5% of Logging two or more occasions/day participants achieving weight losses ≥ 5% from baseline compared to 71.2% of Logging one occasion/day participants, 55.6% of Backlogging participants, and 19.8% of Minimal Logging participants (all between‐group ps < 0.05).
TABLE 3.
Differences in percent weight loss between clusters.
| Clusters a | Mean diff b | Std error | df | t | p |
|---|---|---|---|---|---|
| 1 vs. 2 | 3.61 | 0.54 | 222.89 | 7.13 | < 0.001 |
| 2 vs. 3 | 2.05 | 0.64 | 102.21 | 2.87 | 0.025 |
| 3 vs. 4 | 1.34 | 0.61 | 91.09 | 1.94 | 0.219 |
| 1 vs. 3 | 5.66 | 0.70 | 94.75 | 8.07 | < 0.001 |
| 1 vs. 4 | 6.99 | 0.47 | 238.16 | 14.97 | < 0.001 |
| 2 vs. 4 | 3.39 | 0.49 | 262.85 | 6.94 | < 0.001 |
Cluster 1 = Minimal Logging; Cluster 2 = Backlogging; Cluster 3 = Logging one occasion/day; Cluster 4 = Logging two or more occasions/day.
Higher numbers indicate less weight lost during the BWLP.
Table 4 provides the demographic and baseline characteristics of participants by cluster. There were no differences between clusters in terms of baseline weight, BMI, gender, race, or ethnicity; however, there were significant differences between clusters in terms of age, education, and income. Bonferroni‐adjusted post hoc tests demonstrated that Minimal Logging participants were younger than Logging one occasion/day, p = 0.002, and Logging two or more occasions/day participants, p = 0.015. Fewer Minimal Logging participants had a college degree or higher compared to Backlogging, p = 0.023, and Logging two or more occasions/day participants, p < 0.001; moreover, Logging one occasion/day also had fewer individuals with a completed college degree or higher than Logging two or more occasions/day, p = 0.043. Finally, Logging two or more occasions/day participants reported significantly higher household income than Minimal Logging, p < 0.001, and Backlogging p = 0.023. Post hoc group comparisons not described were not statistically significant, all ps > 0.05.
TABLE 4.
Participant demographic and baseline characteristics by cluster.
| Variable | Minimal logging (n = 101) | Backlogging (n = 124) | Logging one occasion/day (n = 59) | Logging two or more occasions/day (n = 165) | p |
|---|---|---|---|---|---|
| Age, years, mean (SD) | 45.9 (10.9) | 49.7 (11.4) | 52.6 (12.0) | 50.2 (10.9) | 0.002 |
| Baseline BMI, kg/m2, mean (SD) | 35.8 (3.6) | 35.8 (4.3) | 35.9 (3.7) | 35.5 (4.2) | 0.892 |
| Baseline weight, kg, mean (SD) | 100.2 (15.8) | 100.3 (15.6) | 101.6 (15.1) | 98.8 (14.6) | 0.634 |
| Gender, n (%) | 0.374 | ||||
| Woman | 84 (83.2) | 99 (79.8) | 48 (81.4) | 144 (87.3) | |
| Man | 17 (16.8) | 25 (20.2) | 11 (18.6) | 21 (12.7) | |
| Race, n (%) | |||||
| American Indian/Alaskan native | 0 (0.0) | 2 (1.6) | 2 (3.4) | 3 (1.8) | 0.369 |
| Asian | 5 (4.9) | 4 (3.2) | 2 (3.4) | 3 (1.8) | 0.516 |
| Black or African American | 21 (20.7) | 32 (25.8) | 18 (30.5) | 34 (20.6) | 0.367 |
| Native Hawaiian/Pacific Islander | 1 (1.0) | 0 (0.0) | 0 (0.0) | 1 (0.6) | 0.797 |
| White | 76 (75.2) | 88 (71.0) | 40 (67.8) | 129 (78.2) | 0.339 |
| Other | 0 (0.0) | 2 (1.6) | 1 (1.7) | 2 (1.2) | 0.610 |
| Ethnicity, n (%) | 0.327 | ||||
| Hispanic or Latino | 10 (9.9) | 17 (13.7) | 5 (8.5) | 12 (7.3) | |
| Not Hispanic or Latino | 91 (90.1) | 107 (86.3) | 54 (91.5) | 153 (92.7) | |
| Level of education, n (%) | 0.006 | ||||
| < College degree | 44 (43.5) | 37 (29.8) | 22 (37.3) | 39 (23.6) | |
| ≥ College degree | 57 (56.4) | 87 (70.2) | 37 (62.7) | 126 (76.4) | |
| Annual household income, USD, n (%) a | 0.007 | ||||
| ≤ $75,000 | 57 (58.2) | 62 (50.4) | 25 (43.9) | 59 (36.9) | |
| > $75,000 | 41 (41.8) | 61 (49.6) | 32 (56.1) | 101 (63.1) | |
Income data not reported for n = 3 participants in minimal logging, n = 1 in backlogging, n = 2 in logging one occasion/day, and n = 5 in Logging two or more occasions/day.
4. Discussion
The current study aimed to characterize within‐day patterns of dietary self‐monitoring within the context of a BWLP. Consistent with the broader literature [2, 5, 6, 8], results demonstrated that greater adherence to self‐monitoring of dietary intake was associated with greater weight loss during a BWLP and that adherence to self‐monitoring decreased over the course of the intervention. As a novel contribution, results quantified within‐day patterns of self‐monitoring; the most frequent pattern of self‐monitoring observed was logging on two or more occasions/day (30.5% of participant days), followed by backlogging (29.2%), not logging at all (20.8%), logging on one occasion/day (17.7%), and pre‐logging (1.8%). Participants were grouped into clusters based on their patterns of self‐monitoring across the BWLP; those who were categorized into the Logging two or more occasions/day and Logging on one occasion/day clusters lost significantly more weight than those in Backlogging and Minimal logging clusters. Moreover, the proportion of participants achieving clinically‐meaningful weight losses was significantly different between all clusters; this proportion was highest in the Logging two or more occasions/day cluster, at 89%, versus 71% in Logging one occasion/day, 56% in Backlogging, and 20% in Minimal logging. Finally, although there were no differences between clusters in terms of BMI, gender, or race/ethnicity, Logging two or more occasions/day participants were older, most likely to have a college degree, and most likely to report household income > 75,000/year. Conversely, Minimal logging participants were the youngest and this cluster had the highest proportion of individuals without a college degree and with household incomes of 75,000/year or less.
These results have important implications for future BWLPs. First, they provide additional support for the importance of recommendations that individuals self‐monitor dietary intake every day. Results also suggest some additional benefit for logging on two or more occasions/day, supporting recommendations for BWLP participants to self‐monitor food/drink consumption throughout the day. Although there was not a statistically significant difference in weight loss between Logging two or more occasions/day and Logging one occasion/day participants (who experienced weight losses of 9.1% and 7.7%, respectively), a greater proportion of Logging two or more occasions/day participants lost a clinically‐meaningful amount of weight during the BWLP. These results are consistent with those of Harvey et al. [8], who concluded that self‐monitoring needed to occur at least two to three times per day (vs. in a single sitting) to promote successful weight loss.
Results also suggest some benefit to Backlogging. It has been less clear whether BWLP participants should be encouraged to backlog missed days or whether they should be encouraged to simply “start over” by beginning to self‐monitor new food/drink consumption after days that self‐monitoring did not occur. Results from the current study suggest that interventionists may want to consider suggesting the former; it may be that, under a “something is better than nothing” approach, participants may still be able to gain insight and awareness into their eating habits from backlogging past eating episodes that they could then apply to future situations.
Finally, participants in the Logging two or more occasions/day group were significantly older, were more likely to have a college degree or higher, and to have a household income over 75,000/year compared to Minimal logging participants. Future research should investigate potential mechanisms of this association (e.g., in relation to free time, stress, and/or caregiving roles) to inform the development of future interventions to support self‐monitoring adherence in younger individuals and those with lower socioeconomic status. For example, there may be promise in implementing less‐burdensome methods of self‐monitoring, such as self‐monitoring only certain meals or on certain days, or using strategies such as the checking boxes or recording counts of certain categories of foods [23]. Indeed, a study that compared a simplified version of self‐monitoring dietary intake (wherein participants were only asked to monitor “red‐zone” foods from the Traffic Lite Diet) to a traditional dietary self‐monitoring app found that the simplified self‐monitoring approach led to higher engagement with self‐monitoring along with higher ratings of satisfaction with and helpfulness of self‐monitoring dietary intake [24].
Taken together, the results also highlight other areas for future intervention development. Encouragingly, the most frequent pattern of self‐monitoring documented in this study was logging on two or more occasions/day (occurring on 30.5% of participant days). Given that the next most frequent patterns were backlogging and not logging at all, future interventions should aim to develop strategies to help support adherence to self‐monitoring of dietary intake. For example, most BWLPs now encourage participants to self‐monitor dietary intake using smartphone apps versus paper‐and‐pencil records; as individuals often carry their smartphones around with them throughout the day, there is opportunity to harness notifications and reminders to encourage self‐monitoring [25]. Finding opportune times to deliver reminders may be especially important; for example, a recent study found that, compared to time‐based notifications (e.g., reminding someone to self‐monitor an hour after their normal lunch time), context‐dependent notifications (i.e., sending reminder notifications for people to self‐monitor at times that the person may be “interruptible” and thus more receptive and able to act on notifications) led to faster response times for clicking the notification and improved the likelihood that a participant would self‐monitor within 60 min of receiving the notification [26].
Results should be interpreted in light of study limitations. Importantly, this was a retrospective study using data from an existing BWLP that gave the same recommendations to self‐monitor dietary intake to all participants. As participants were not randomized to different patterns of self‐monitoring, caution should be taken with causal interpretation of results. It is possible that outside factors influenced both participants' patterns of self‐monitoring and their weight loss outcomes (e.g., if having a busy schedule or caregiving needs affected both the time that a participant had to engage in self‐monitoring of dietary intake and their food choices). It is also possible that the patterns of engagement in self‐monitoring observed in this study (and thus our clustering of participants by these patterns) may more broadly represent the level of motivation to engage in the weight loss program and program goals. The current study's ability to accurately document “pre‐logging” was also limited due to the way that study data were collected and coded. As foods/drinks that were pre‐logged but later edited would be over‐written with a new time‐stamp, it is possible that participants were pre‐logging but that this day would be coded as another category if updated on the day of consumption or afterward. Therefore, the prevalence of pre‐logging captured in the current study is likely an under‐estimate, precluding the ability to investigate associations between this pattern and weight‐loss outcomes. The pre‐logging category is also unlikely to fully capture the behavior of pre‐planning, as individuals may pre‐plan (and pre‐log) earlier on the same day of consumption (or, conversely, they may pre‐plan without pre‐logging). The prevalence of backlogging may similarly be over‐estimated due to data capture and coding methods, as a day that an individual self‐monitored dietary intake throughout the day but changed one of these entries on a later day would have been categorized as backlogging. Moreover, the current study only approximated capture of “real‐time logging” by coding whether logging occurred on two or more occasions (an hour or more apart) each day as the data capture methods precluded the ability to assess the proximity of logging to food/drink consumption. Specifically, the data collected by FatSecret included the meal at which foods/drinks were consumed (e.g., breakfast, lunch, dinner, or snack) but not the time that consumption or self‐monitoring occurred (and meal times may vary by individual or due to travel outside of one's usual time zone). To replicate and extend the results of the current study, future studies should be conducted that collect more accurate data both on when foods/drinks were consumed and when these items were logged. For example, methods used by Burke et al. [6] could be updated for use with newer self‐monitoring technologies, with participants asked what times they consumed foods/drinks and timestamps automatically recorded when items are tracked in the app (with changes logged but not in a way that overwrites the initial entries). Finally, the current study did not use thresholds (in relation to number of items or kcal consumed/day) to determine the “completeness” of self‐monitoring; thus, future studies should examine whether there is an interaction between complete versus partial self‐monitoring (e.g., self‐monitoring only a few of the foods/drinks consumed in a day) and within‐day patterns of self‐monitoring in relation to weight change.
Despite these limitations, the current study has several important strengths, including the collection of objective, time‐stamped self‐monitoring data from a large sample of individuals who participated in a gold‐standard BWLP. Moreover, analyses used an intent‐to‐treat approach, such that all participants who enrolled in the initial intervention were used in analyses; it was conservatively assumed that participants who dropped out of the study regained 0.3 kg/month, an assumption used in several prior BWLP trials [17, 18].
Taken together, the results support existing recommendations [6] for BWLP participants to log daily, and preferably throughout the day as consumption occurs. Results also highlighted important patterns in relation to equity in BWLP outcomes, as individuals who engaged in Logging two or more occasions/day were older, more likely to have a college degree or higher educational level, and to have incomes > 75,000/year. Future studies should replicate these findings and use randomized study designs that can allow for investigation into the causality of associations demonstrated between more frequent self‐monitoring and greater success at weight loss.
Funding
Support for the current study was provided by the National Institute of Health, National Institute of Diabetes Digestive and Kidney Diseases (Grant R01DK119244).
Disclosure
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Wadden T. A., Tronieri J. S., and Butryn M. L., “Lifestyle Modification Approaches for the Treatment of Obesity in Adults,” American Psychologist 75, no. 2 (2020): 235–251, 10.1037/amp0000517. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Yu Z., Sealey‐Potts C., and Rodriguez J., “Dietary Self‐Monitoring in Weight Management: Current Evidence on Efficacy and Adherence,” Journal of the Academy of Nutrition and Dietetics 115, no. 12 (2015): 1931–1938, 10.1016/j.jand.2015.04.005. [DOI] [PubMed] [Google Scholar]
- 3. Carpenter C. A., Ugwoaba U. A., Cardel M. I., and Ross K. M., “Using Self‐Monitoring Technology for Nutritional Counseling and Weight Management,” Digital Health 8 (2022): 20552076221102774, 10.1177/20552076221102774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kanfer F. H., “Self‐Monitoring: Methodological Limitations and Clinical Applications,” Journal of Consulting and Clinical Psychology 35, no. 2 (1970): 148–152, 10.1037/h0029874. [DOI] [Google Scholar]
- 5. Burke L. E., Wang J., and Sevick M. A., “Self‐Monitoring in Weight Loss: A Systematic Review of the Literature,” Journal of the American Dietetic Association 111, no. 1 (2011): 92–102, 10.1016/j.jada.2010.10.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Burke L. E., Sereika S. M., Music E., Warziski M., Styn M. A., and Stone A., “Using Instrumented Paper Diaries to Document Self‐Monitoring Patterns in Weight‐Loss,” Contemporary Clinical Trials 29, no. 2 (2008): 182–193, 10.1016/j.cct.2007.07.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Baker R. C. and Kirschenbaum D. S., “Self‐Monitoring May be Necessary for Successful Weight Control,” Behavior Therapy 24, no. 3 (1993): 377–394, 10.1016/s0005-7894(05)80212-6. [DOI] [Google Scholar]
- 8. Harvey J., Krukowski R., Priest J., and West D., “Log Often, Lose More: Electronic Dietary Self‐Monitoring for Weight Loss,” Obesity 27, no. 3 (2019): 380–384, 10.1002/oby.22382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ross K. M., Qiu P., You L., and Wing R. R., “Week‐to‐Week Predictors of Weight Loss and Regain,” Health Psychology 38, no. 12 (2019): 1150–1158, 10.1037/hea0000798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Peterson N. D., Middleton K. R., Nackers L. M., Medina K. E., Milsom V. A., and Perri M. G., “Dietary Self‐Monitoring and Long‐Term Success With Weight Management,” Obesity 22, no. 9 (2014): 1962–1967, 10.1002/oby.20807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Burke L. E., Warziski M., Starrett T., et al., “Self‐Monitoring Dietary Intake: Current and Future Practices,” Journal of Renal Nutrition 15, no. 3 (2005): 281–290, 10.1016/j.jrn.2005.04.002. [DOI] [PubMed] [Google Scholar]
- 12. Tang H. B., Jalil N. I. B. A., Tan C. S., He L., and Zhang S. J., “Why More Successful? An Analysis of Participants’ Self‐Monitoring Data in an Online Weight Loss Intervention,” BMC Public Health 24, no. 1 (2024): 322, 10.1186/s12889-024-17848-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hayes J. F., Balantekin K. N., Fitzsimmons‐Craft E. E., et al., “Greater Average Meal Planning Frequency Predicts Greater Weight Loss Outcomes in a Worksite‐Based Behavioral Weight Loss Program,” Annals of Behavioral Medicine 55, no. 1 (2020): 14–23, 10.1093/abm/kaaa021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ross K. M., Shankar M. N., Qiu P., et al., “Adaptive vs Monthly Support for Weight‐Loss Maintenance: A Randomized Clinical Trial,” JAMA Network Open 8, no. 9 (2025): e2532681, 10.1001/jamanetworkopen.2025.32681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ross K. M., Shankar M. N., Qiu P., et al., “Design of Project STAR: A Randomized Controlled Trial Evaluating the Impact of an Adaptive Intervention on Long‐Term Weight‐Loss Maintenance,” Contemporary Clinical Trials 146 (2024): 107707, 10.1016/j.cct.2024.107707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Diabetes Prevention Program (DPP) Research Group , “The Diabetes Prevention Program (DPP): Description of Lifestyle Intervention,” Diabetes Care 25, no. 12 (2002): 2165–2171, 10.2337/diacare.25.12.2165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Perri M. G., Shankar M. N., Daniels M. J., et al., “Effect of Telehealth Extended Care for Maintenance of Weight Loss in Rural US Communities: A Randomized Clinical Trial,” JAMA Network Open 3, no. 6 (2020): e206764, 10.1001/jamanetworkopen.2020.6764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Wing R. R., Tate D. F., Gorin A. A., Raynor H. A., and Fava J. L., “A Self‐Regulation Program for Maintenance of Weight Loss,” New England Journal of Medicine 355, no. 15 (2006): 1563–1571, 10.1056/NEJMoa061883. [DOI] [PubMed] [Google Scholar]
- 19. Ross K. M. and Wing R. R., “Concordance of In‐Home ‘Smart’ Scale Measurement With Body Weight Measured In‐Person,” Obesity Science & Practice 2, no. 2 (2016): 224–248, 10.1002/osp4.41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Pebley K., Klesges R. C., Talcott G. W., Kocak M., and Krukowski R. A., “Measurement Equivalence of E‐Scale and In‐Person Clinic Weights,” Obesity (Silver Spring) 27, no. 7 (2019): 1107–1114, 10.1002/oby.22512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Krukowski R. A. and Ross K. M., “Measuring Weight With Electronic Scales in Clinical and Research Settings During the Coronavirus Disease 2019 Pandemic,” Obesity 28, no. 7 (2020): 1182–1183, 10.1002/oby.22851. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. FatSecret , fatsecretPlatform API (2025), https://platform.fatsecret.com/docs/guides.
- 23. Krukowski R. A., Harvey J., Borden J., Stansbury M. L., and West D. S., “Expert Opinions on Reducing Dietary Self‐Monitoring Burden and Maintaining Efficacy in Weight Loss Programs: A Delphi Study,” Obesity Science & Practice 8, no. 4 (2022): 401–410, 10.1002/osp4.586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Patel M. L., Cleare A. E., Smith C. M., Rosas L. G., and King A. C., “Detailed Versus Simplified Dietary Self‐Monitoring in a Digital Weight Loss Intervention Among Racial and Ethnic Minority Adults: Fully Remote, Randomized Pilot Study,” JMIR Formative Research 6, no. 12 (2022): e42191, 10.2196/42191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Woodward J., Chen Y. P., Jurczyk K., Ross K. M., Anthony L., and Ruiz J., “A Survey of Notification Designs in Commercial mHealth Apps,” Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, ACM, 2021, 1–7, 10.1145/3411763.3451639. [DOI] [Google Scholar]
- 26. Chen Y. P., Woodward J., Bista D., et al., “Investigating Contextual Notifications to Drive Self‐Monitoring in mHealth Apps for Weight Maintenance,” Proceedings of the CHI Conference on Human Factors in Computing Systems, ACM, 2024, 1–21, 10.1145/3613904.3641993. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.