This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Mobile apps for weight loss provide users with convenient features for recording lifestyle and health indicators; they have been widely used for weight loss recently. Previous studies in this field generally focused on the relationship between the cumulative nature of self-reported data and the results in weight loss at the end of the diet period. Therefore, we conducted an in-depth study to explore the relationships between adherence to self-reporting and weight loss outcomes during the weight reduction process.
We explored the relationship between adherence to self-reporting and weight loss outcomes during the time series weight reduction process with the following 3 research questions: “How does adherence to self-reporting of body weight and meal history change over time?”, “How do weight loss outcomes depend on weight changes over time?”, and “How does adherence to the weight loss intervention change over time by gender?”
We analyzed self-reported data collected weekly for 16 weeks (January 2017 to March 2018) from 684 Korean men and women who participated in a mobile weight loss intervention program provided by a mobile diet app called Noom. Analysis of variance (ANOVA) and chi-squared tests were employed to determine whether the baseline characteristics among the groups of weight loss results were different. Based on the ANOVA results and slope analysis of the trend indicating participant behavior along the time axis, we explored the relationship between adherence to self-reporting and weight loss results.
Adherence to self-reporting levels decreased over time, as previous studies have found. BMI change patterns (ie, absolute BMI values and change in BMI values within a week) changed over time and were characterized in 3 time series periods. The relationships between the weight loss outcome and both meal history and self-reporting patterns were gender-dependent. There was no statistical association between adherence to self-reporting and weight loss outcomes in the male participants.
Although mobile technology has increased the convenience of self-reporting when dieting, it should be noted that technology itself is not the essence of weight loss. The in-depth understanding of the relationship between adherence to self-reporting and weight loss outcome found in this study may contribute to the development of better weight loss interventions in mobile environments.
Mobile apps are widely used for tracking weight loss [
Self-reporting has been recognized as a very important behavioral treatment method for weight loss [
Compared to traditional tools, mobile technology’s portability has made it easier and more effective to monitor and control one’s weight [
However, an in-depth understanding is lacking regarding the relationship between weight loss outcomes and the dynamics of adherence to self-reporting according to specific groups. Attempts at an in-depth understanding of this relationship is very important for several reasons. First, given that demographic characteristics affect both adherence to self-reporting and dietary approaches [17–22], researchers may gain a fundamental understanding of how certain groups respond to weight loss interventions. Second, researchers may understand the mechanism of weight changes during the weight loss process that involves a series of weight gain, loss, and regain cycles [
Thus, this study aimed to analyze the relationship between weight loss outcomes and the dynamics of adherence to self-reporting. To that end, we analyzed self-reported data that were aggregated on a weekly basis for 16 weeks, from January 2017 to March 2018, from 684 Korean men and women who participated in a mobile weight loss intervention program provided by a commercial smartphone app.
The following 3 research questions were explored and answered: How does adherence to self-reporting on body weight and meal history change over time? How do weight loss outcomes depend on weight changes over time? How does adherence to the weight loss intervention change over time by gender?
The mobile intervention program for this study was delivered through a commercially available mobile coaching program called Noom [
All coaches were registered dietitians, and they were supervised weekly by a clinical psychologist. They helped users set weekly goals and provided personalized feedback based on each user’s lifestyle. Messages were sent to the users at least twice per week.
The frequency of weight entries was analyzed. In most health care app research programs, user adherence to a certain intervention has often been measured by the frequency of self-reporting [
The study participants reported their weight through the app. After reporting, weight-related menu-usage logs were automatically stored on a server. The collected usage log data were summed at weekly intervals to generate data indicating adherence to weight-reporting activity over 16 weeks, for each subject.
BMI is often used to determine whether a person is within a healthy weight range based on their height [
Participants reported their body weight daily through Noom. The daily BMI values for each participant were calculated by dividing the daily reported weights by their initially entered height values, and the daily results were averaged on a weekly basis.
BMI delta values were used to identify weekly weight fluctuation trends [
As with weight entries, the frequency of self-reported meal history entries has frequently been used in prior studies to address the level of adherence to weight loss intervention programs [
These food types were categorized as green, yellow, and red groups based on their caloric density, that is, calories per gram or milliliter [
Variables representing food intake by food type were generated based on the aforementioned criteria. Furthermore, food intake was reported 6 times per day (ie, breakfast, morning snacks, lunch, afternoon snacks, dinner, and evening snacks). This provided a substantial amount of data regarding the changes in food intake for the 3 food groups of each participant over 16 weeks.
This study was approved by the Institutional Review Board of the Asan Medical Center, Korea (No. 2017-1253).
Study sample selection process.
Participants measured their weight on their scales and then reported the weight via the app. To prevent incorrect input that may occur when users report these weight values, certain reported values were corrected. As no previous studies provided any clear criteria for mitigating data errors, a group of three experts consisting of doctors and health care professionals set the error criteria and reviewed the data. It was assumed that there could be potential errors if the difference between the maximum weight and minimum weight per week was ≥3 kg. The daily weight and BMI values of 101 cases corresponding to this assumption were not included in the weekly averages.
To analyze the self-reported time-series pattern based on weight loss groups, subjects were divided into 3 groups according to recommendations from prior research [
Analysis of variance (ANOVA) and chi-squared tests were performed to analyze whether the baseline characteristics of the 3 groups were statistically different. Furthermore, ANOVAs along the time axis were performed to determine whether each measurement (ie, BMI, weight entry, and meal history entry trends) of the 3 groups (ie, <5%, 5%-10%, >10%) within gender were statistically different over 16 weeks.
Participant characteristics for the total sample and according to outcome group.
Characteristics | Total sample | Outcome group according to level of weight loss | ||||
<5% | 5%-10% | >10% | ||||
|
|
|
|
|
|
|
|
Male | 218 (31.9)a | 126 | 73 | 19 | .39b |
|
Female | 466 (68.1)a | 294 | 133 | 39 | |
Age (years), mean (SD) | 33.31 (7.6) | 33.7 | 32.87 | 32.09 | .19c | |
|
|
|
|
|
|
|
|
Obese | 178 (26)a | 112 | 49 | 17 | .62b |
|
Overweight | 506 (74)a | 308 | 157 | 41 |
an (%).
bTested differences between the 3 groups using chi-squared tests.
cTested differences between the 3 groups using analysis of variance (ANOVA).
The number of weight entries on a weekly basis over 16 weeks; the colored bands show the 95% CI.
Significance of analysis of variance (ANOVA) test results comparing the differences between groups for 16 weeks.
Group | Week | |||||||||||||||||||||||||||||||
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | |||||||||||||||||
|
||||||||||||||||||||||||||||||||
|
Totala | **b | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | **b | ***c | |||||||||||||||
|
Mena |
|
|
|
**b | **b |
|
|
|
*d |
|
|
||||||||||||||||||||
|
Womena | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c |
|
***c | |||||||||||||||
|
||||||||||||||||||||||||||||||||
|
Totala | *d | *d | **b | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | ***c | |||||||||||||||||||
|
Mena |
|
|
|
*d | *d |
|
|
**b | **b | **b | **b | *d | |||||||||||||||||||
|
Womena | *d | ***c | **b | **b | ***c | ***c | ***c | ***c | ***c | ***c | |||||||||||||||||||||
|
||||||||||||||||||||||||||||||||
|
Totala | ***c | ***c | ***c | *d | ***c | ***c | *d | **b | *d | ***c | *d | ||||||||||||||||||||
|
Mena | **b |
|
|
|
|
|
|
|
|
|
|
|
|||||||||||||||||||
|
Womena | ***c | ***c | **b |
|
***c | ***c | *d | *d |
|
|
***c | **b | |||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
||||||||||||||||||||
|
Totala | *d | *d | **b | *d | ***c | ***c | ***c | ***c | ***c | **b | ***c | *d | **b | *d | **b | ||||||||||||||||
|
Mena | |||||||||||||||||||||||||||||||
|
Womena | *d | *d | **b | ***c | **b | **b | ***c |
|
***c | *d | **b | **b | **b |
aWhen the null hypotheses were tested using ANOVA, the three groups (weight loss >10%, weight loss of 5%-10%, and weight loss <5%) had the same value.
b
c
d
eDefined as the difference between the maximum and minimum BMIs in a given week.
In all 3 groups, the BMI delta values — representing the degree of weight fluctuation — were only high during the initial 3 weeks (
BMIs over 16 weeks according to weight loss outcome groups; the colored bands show the 95% CI.
BMI delta over 16 weeks according to weight loss outcome groups; the colored bands show the 95% CI.
As seen in
Number of meal history entries on a weekly basis over 16 weeks; the colored bands show the 95% CI.
Significance of analysis of variance (ANOVA) test results comparing the differences in intake amount by food group (green, yellow, and red) in women for 16 weeks.
Food group | Week | Avga | |||||||||||||||||
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 |
|
|||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
BREb |
|
|
**c |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MOR.Sd |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
LUNe |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
AFT.Sf |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DINg |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
EVE.Sh |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
BRE |
|
|
|
|
|
|
|
|
|
|
**c |
|
|
|
|
|
***i | |
|
MOR.S |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
LUN |
|
|
|
**c |
|
|
|
|
**c |
|
**c | **c |
|
**c |
|
|
***i | |
|
AFT.S |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DIN | **c |
|
|
*j |
|
**c | **c |
|
*j |
|
*j | **c |
|
|
**c |
|
***i | |
|
EVE.S |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||
|
BRE |
|
|
|
**c | *j |
|
|
|
|
*j | **c |
|
|
|
|
|
***i | |
|
MOR.S |
|
|
|
|
|
**c |
|
|
*j |
|
|
|
|
|
|
|
**c | |
|
LUN |
|
|
**c | **c | *j | **c | *j | **c | *j | *j | *j | **c |
|
|
|
**c | ***i | |
|
AFT.S |
|
**c | *j |
|
|
|
|
|
|
|
|
**c |
|
|
|
|
***i | |
|
DIN |
|
**c | **c | **c | **c |
|
|
*j | *j | **c | **c | *j | **c |
|
**c |
|
***i | |
|
EVE.S | *j |
|
|
|
|
|
*j |
|
|
**c |
|
|
|
|
|
|
***i |
aAvg: grouped average of food intake over 16 weeks; the
bBRE: breakfast.
c
dMOR.S: morning snack.
eLUN: lunch.
fAFT.S: afternoon snack.
gDIN: dinner.
hEVE.S: evening snack.
i
j
Previous studies showed that the degree of adherence to self-reporting in mobile weight loss intervention programs decreases over time [
Among the many factors that hinder long-term adherence, the most relevant to this study may be the lack of serious awareness of current health conditions [
Conversely, it may be entirely reasonable to simply admit that humans are less committed to interventions over time, especially for slow and arduous tasks like trying to lose weight. In other words, it may be wise to avoid long-term adherence issues that are unlikely to be solved. Rather, the best practice may be to focus on understanding the characteristics of adherence with latent patterns. Recent data science algorithms may be employed to understand specific patterns of adherence. Specifically, the use of time series decomposition algorithms that eliminate the inevitable long-term declines in adherence may be used for exploring repeated patterns of adherence [
According to the results of this empirical analysis, BMI changes can be characterized in 3 time series periods (
The second interval is between weeks 3 and 13. This interval was characterized by differences in BMI that increased over time among the outcome groups. Further, most BMI deltas were statistically different among the groups (
The last interval is between weeks 13 and 16. This interval was characterized by no differences in BMI deltas between the groups (
Gender effects on the differences in self-reporting adherence have often been discussed in previous studies of weight loss interventions [
Unlike female participants, male participants had little differences in the number of weight entries among the outcome groups during the 16 weeks (
These results suggest that there was no statistical association between adherence to self-reporting and weight loss outcomes in the male population. This contradicts some previous studies that showed positive relationships between the two [
Another interesting finding in this analysis is that the male participants, unlike the female participants, demonstrated no statistical difference in the intake of food types among the 3 weight loss groups. Previous studies found that men were insensitive to food intake during their diets, while women were very sensitive [
This study has limitations that may inspire future research. First, analysis results cannot be free from errors, as the data are self-reported. Particularly, the weight statistics used to divide participants into 3 weight loss groups were self-reported. Although reporting of actual information was encouraged in the research agreements, potential mistakes could have arisen for which participants could have been unaware while reporting their weight. Furthermore, scales that the participants used may not have been reliable. To prevent these potential problems, some weight values were modified by applying logic during preprocessing. However, the applied logic was empirical and somewhat arbitrary. In addition, all reported data other than weight information, such as gender and age, were assumed to be true and consistent throughout the study. No correction was made for misreported data. Further research should be conducted in more sophisticated environments by collecting data under rigorous verification by researchers or health professionals.
Second, there is a potential limitation related to the approach and data analyses in this study. First, the significance of the
Third, there may be potential limitations in the experimental setting assumptions. For instance, although the registered dietitians in the program did not recommend the use of weight-reducing drugs like diuretics, some participants might have relied on some medicines that have a huge effect on weight loss. Furthermore, the study design did not include a control group, which might have undermined the ability to evaluate the efficacy of the mobile weight loss intervention. For example, the weights of the subjects who canceled the program prior to week 13 were not analyzed. By analyzing such data in a control group, the efficacy of the mobile intervention can be evaluated in a more rigorous manner. In addition, significant differences in food intake reporting by weight loss outcomes in female participants may have been caused by the Hawthorne effect [
Fourth, two factors may have undermined the representativeness of self-reported data for measurements. First, the number of self-reported entries may not have measured spontaneous adherence levels, because the app provides push notifications to remind users to report their data. Additionally, as the degree of self-reporting decreases over time, the data may be insufficient for representing weight status. Particularly, the average number of weight entries in the 16th week for the lowest weight loss group (<5%) was less than 1.5 per participant. When body weight is reported only once per week, potentially biased values may have been reported depending on the timing of the report (eg, after exercise or immediately after main meals, mornings versus evenings). Therefore, for studies requiring self-reporting, further discussion is required regarding data collection and analytical approaches.
Mobile technology has increased the convenience of self-reporting when dieting [
Food-type criteria according to caloric density.
P values from the ANOVA tests comparing the differences between groups for 16 weeks.
P values from the ANOVA tests comparing the differences in intake amount by food group (green, yellow, and red) in women for 16 weeks.
analysis of variance.
This study was supported by a grant from the Research and Development Project, Ministry of Trade, Industry, and Energy, Republic of Korea (no. 20004503).
YK is employed at Noom and has conflicts of interest. All other authors declare no conflicts of interest.