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 JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included.
Although calorie tracking is one of the strongest predictors of weight loss in behavioral weight loss interventions, low rates of adherence are common.
This study aims to examine the feasibility and acceptability of using the Slip Buddy app during a 12-week web-based weight loss program.
We conducted a randomized pilot trial to evaluate the feasibility and acceptability of using the Slip Buddy app compared with a popular commercial calorie tracking app during a counselor-led, web-based behavioral weight loss intervention. Adults who were overweight or obese were recruited on the web and randomized into a 12-week web-based weight loss intervention that included either the Slip Buddy app or a commercial calorie tracking app. Feasibility outcomes included retention, app use, usability, slips reported, and contextual factors reported at slips. Acceptability outcomes included ratings of how helpful, tedious, taxing, time consuming, and burdensome using the assigned app was. We described weight change from baseline to 12 weeks in both groups as an exploratory outcome. Participants using the Slip Buddy app provided feedback on how to improve it during the postintervention focus groups.
A total of 75% (48/64) of the participants were female and, on average, 39.8 (SD 11.0) years old with a mean BMI of 34.2 (SD 4.9) kg/m2. Retention was high in both conditions, with 97% (31/32) retained in the Slip Buddy condition and 94% (30/32) retained in the calorie tracking condition. On average, participants used the Slip Buddy app on 53.8% (SD 31.3%) of days, which was not significantly different from those using the calorie tracking app (mean 57.5%, SD 28.4% of days), and participants who recorded slips (30/32, 94%) logged on average 17.9 (SD 14.4) slips in 12 weeks. The most common slips occurred during snack times (220/538, 40.9%). Slips most often occurred at home (297/538, 55.2%), while working (153/538, 28.4%), while socializing (130/538, 24.2%), or during screen time (123/538, 22.9%). The conditions did not differ in participants’ ratings of how their assigned app was tedious, taxing, or time consuming (all values of
Self-monitoring of dietary lapses and the contextual factors associated with them may be an alternative for people who do not prefer calorie tracking. Future research should examine patient characteristics associated with adherence to different forms of dietary self-monitoring.
ClinicalTrials.gov NCT02615171; https://clinicaltrials.gov/ct2/show/NCT02615171
Obesity is a major risk factor for type 2 diabetes [
Interestingly, calorie tracking is a relatively complex form of self-monitoring compared with forms used for other behaviors. For example, in smoking cessation interventions, the smoker keeps a tally of the number of cigarettes smoked each day and notes the triggers associated with each smoking episode [
Previous ecological momentary assessment studies have revealed that people are able to identify and self-monitor dietary lapses [
Employing a user-centered design, we developed the Slip Buddy app, which allows users to track dietary lapses as they occur and the contextual factors that cued each lapse [
This study is a pilot feasibility randomized controlled trial in which participants with overweight or obesity were randomized to receive either the Slip Buddy app or a commercial calorie tracking app during a 12-week counselor-led web-based weight loss intervention. Our first aim is to compare groups on retention and app use to explore whether the commercial app is significantly superior to Slip Buddy, which would point to the need for further modification to Slip Buddy before proceeding to a fully powered efficacy trial. Our second aim is to describe the total number of slips reported and the contextual factors reported at slips, including the location of slips, type of eating episode (eg, lunch and snack), stress, and hunger/satiety. This aim is descriptive in nature. Our third aim is to assess the usability, acceptability, and burden of the Slip Buddy app quantitatively and via qualitative interviews where participants shared what they liked and disliked about the app and the features they would like added. Similar to aim 1, we tested whether the commercial app was significantly superior in terms of usability, acceptability, and burden, which would signal areas for further modification to Slip Buddy before proceeding to an efficacy trial. Our fourth aim is to describe the percentage of weight loss from baseline to 12 weeks in both groups, including the proportion of participants who lost clinically significant weight. This aim is exploratory because only a fully powered efficacy trial could address this question.
We conducted a pilot feasibility randomized trial in which participants who were overweight or obese were recruited into a remotely delivered intervention via web-based advertisements at the University of Connecticut, ResearchMatch, and Facebook groups across the United States between July and October 2019. All work was approved by the University of Connecticut Institutional Review Board. We recruited people interested in losing weight with BMI between 27 and 45 kg/m2, aged 18-65 years, who had an Android smartphone, and who had phone connectivity at home and work. Exclusion criteria were inability to walk unaided for one-fourth mile without stopping, not being a daily Facebook user (because the group-based part of the intervention was delivered via Facebook), taking medications known to affect appetite and/or weight, having a condition that precludes dietary changes (eg, ulcerative colitis), type 1 or 2 diabetes, gastric bypass surgery or plans to do so during the study period, pregnancy or lactation, severe mental illness or substance use disorder, binge eating disorder, or loss of 5% or more weight in the past 3 months. Recruitment ads contained a link to a screening survey that included a study description, initial informed consent, and screening questions. Participants eligible after the screening survey were emailed the consent form and completed a telephone screening call. The screening call included reviewing the consent form, any remaining eligibility-related questions, and an emailed link to the baseline survey.
Before randomization, potential participants were required to attend an orientation webinar in which study staff used a methods-motivational interviewing approach to help them understand the scientific rationale of the trial design, research questions, and methods. This helps participants understand the commitment entailed in trial enrollment and helps set clear expectations (eg, transparency about the length of assessments), explain the scientific rationale for procedures (eg, randomization and feasibility versus efficacy testing), diffuse ambivalence about research participation using motivational interviewing techniques, and make explicit commitments to self and trial methods. Upon completion, those interested in proceeding with the study were mailed a Wi-Fi scale (Fitbit Aria) and asked to provide staff with log-in information for the scale so that weight could be recorded for assessments. After randomization, participants had a 60-minute call with a study staff person to receive guidance on how to download and use their assigned app and enter their assigned Facebook group. Participants were allowed to keep the scale and were compensated for completing the assessments.
Participants in both conditions were assigned one of the diet tracking apps and received the Diabetes Prevention Program (DPP) lifestyle intervention delivered within a counselor-led private Facebook group that included all participants in their respective conditions. Participants randomized to the Slip Buddy condition were provided the Slip Buddy app, and participants randomized to the calorie tracking condition were instructed to install the free, commercially available MyFitnessPal app. Each group had a different counselor who was either a registered dietitian or a clinical psychologist, and each was trained in the app assigned to their respective conditions and led the Facebook group for that condition.
As in our previous work [
As described above, we developed the Slip Buddy app, which assists users in tracking nonhomeostatic eating and the contextual factors surrounding it (
Slip Buddy Screen Shots.
Participants randomized to the calorie tracking condition were instructed to download MyFitnessPal, a free, commercially available mobile app that provides users with a personalized calorie goal and allows them to track their caloric intake and energy expenditure via exercise in an effort to stay within that goal. Participants were asked to enter everything they eat and drink throughout each day and all of their structured physical activity. They were asked to stay within their calorie goals to facilitate a weight loss of 1-2 lb per week. The group counselor instructed participants to use MyFitnessPal daily and to inspect their dietary entries for high-calorie foods that could be eliminated to achieve their calorie goal.
Retention was defined as the percentage of participants in each condition who completed the 12-week follow-up measures, which included weigh-in and a survey.
We report 3 metrics of app use: (1) whether participants used their assigned app at least once during the 12-week intervention, (2) the total number and percentage of possible days participants used their app over the 12 weeks, and (3) whether participants used their assigned app at least once in week 12 (ie, sustained engagement). As the nature of diet tracking differed between the 2 treatment conditions, how we assessed app use also differed. For participants in the Slip Buddy condition, we intended to categorize participants as having used the Slip Buddy app for a given day if (1) backend data from the app revealed at least one slip was recorded or check-in completed (optional) or (2) in the absence of slips, the participants responded to the end-of-day check-in saying that they did not have any slips that day. Staff reviewed the data in the Slip Buddy database server (ie, backend data) and recorded the number of days each week each participant used the app (eg, either recorded a slip or responded to a notification or check-in reporting that they experienced no slips). However, some participants reported that they did not see or receive the end-of-day check-in notifications from the app, which we determined was related to the authentication token on the phone expiring periodically. The end-of-the-day notification gives the participant the opportunity to confirm that no slips occurred if none had been recorded thus far. Without the notification, we could not distinguish between a day in which the participant did not track slips and a day in which no slips occurred. For this reason, backend data would be an underestimation of app use. As failure can sometimes occur while transmitting app data to the remote database server, we also collected self-report app use data by emailing participants each week a single item asking them how many days they used the app to track slips that week. Self-report data were available for 74.4% (282/379) of weeks across all participants (counting only 7 weeks for the participant who withdrew because of pregnancy). As 26% of self-reported data were missing and backend data were incomplete by an unknown amount, we leveraged both forms of data to measure app use. We used the larger of the 2 values for 2 reasons: (1) when self-reported data are higher than backend data, it could correct for the underestimation bias of backend data and (2) when backend data are higher than self-report, it could correct for recall bias from self-report. The weakness is that we do not have a way to correct for recall bias from self-report that overestimates use, which surely exists to some extent. Self-report data were used for 58.1% (220/379 weeks) of the total weeks, and backend data were used for 27.4% (104/379 weeks) of weeks, which includes the 7 weeks when the backend data were higher than the self-report data and the 97 weeks in which self-report data were missing. On the remaining 14.5% (55/379 weeks) of weeks, self-report and backend data were the same so that the value was used. Although there is no way to correct for possible overestimations via self-report, on 25.6% (97/379 weeks) of weeks, only backend data were available, which would be an underestimate for those weeks. As a sensitivity analysis, we calculated app use using self-reported data if available, and when self-report data were not available, backend data were used. These metrics only differ from the main analysis for the 7 weeks (1 week for each of the 7 participants) where app use abstracted from the backend was higher than self-reported app use.
For participants in the calorie tracking condition, research staff reviewed MyFitnessPal records, and we coded a
Using the above definitions, we calculated the number and percentage of days participants in each treatment condition used their assigned app over the 12-week intervention. As the Slip Buddy app was down for 2 days in week 3, participants in this condition could have only used the app on a maximum of 82 days versus the 84 possible days for participants in the calorie tracking condition. We also categorized participants in both conditions as to whether they used their assigned app at least once over the 12-week intervention and whether they used their assigned app during week 12 as a measure of sustained engagement. Two participants were withdrawn or dropped out of the intervention because of incident pregnancies. For these women, app use was not assessed after they were no longer in the intervention (after week 7 for the Slip Buddy participant who became pregnant and after week 2 for the calorie tracking participant); instead, the calculation of percentage of days the app was used only counted days they were in the intervention.
Backend data from the app were used to describe the number of slips reported for each participant during the intervention period and contextual factors related to slips.
Participants were asked to enter a note about where they were when the slip occurred. The first author collapsed responses into categories that included work, home, other person’s house, restaurant/bar, at an event (eg, football game), in the car, or at the gym.
Participants also indicated the nature of the eating episode in which the slip occurred, which included the choices of breakfast, lunch, dinner, dessert, snack, or alcohol. Alcohol was included to capture drinking episodes that happen outside of the context of meals or snacks and to prompt participants to think of excess alcohol intake as a dietary slip.
Participants also indicated what they were doing from a drop-down menu of domestic activities (eg, chores), working/studying (eg, employment and school), socializing, screen time, or commuting.
When they entered a slip, participants rated how much stress they were experiencing before their slip on a scale of 0 to 10, where 0 indicates no stress and 10 indicates extreme stress. Stress scores of 5 and above were considered moderate to high stress, whereas stress ratings of less than 5 were considered low stress. Participants also rated how hungry or full they felt before they slipped on a scale of 0 to 10, where 0 indicates extremely hungry, 5 indicates comfortably full, and 10 indicates stuffed, that is, uncomfortably full.
The participants were asked to record their diet every day for 12 weeks. These data were extracted from MyFitnessPal and coded for analysis. The first level of coding included recording each day that the participant entered at least one item. The second level extracted the number of eating episodes and calories each day.
The System Usability Scale (SUS) [
At 12 weeks, participants in both conditions rated the helpfulness and ease of use of their assigned app (response options: strongly disagree, disagree, neutral, agree, or strongly agree). We dichotomized responses as strongly agree/agree versus strongly disagree/disagree/neutral. As the Slip Buddy app is exclusively focused on diet, unlike calorie tracking apps that address both diet and exercise, we included a follow-up question asking participants to rate whether a feature that would allow them to track exercise slips (ie, times when they had planned to exercise but did not follow through) would increase the effectiveness of Slip Buddy app. Acceptability was also evaluated in postintervention focus groups via 2 questions: “what did you like most about Slip Buddy app and why?” and “what did you like least about Slip Buddy app and why?” During the intervention, participants started a discussion about the possibility of the Slip Buddy app having a feature that would allow people to track when they were tempted to slip but resisted that temptation. Given the enthusiasm for the idea, we added a question to the focus group script asking participants about the extent to which they would like to track temptations that did not turn into slips.
At 12 weeks, participants in both conditions rated how burdensome it was to use their assigned app on a scale of 0 to 100, with 0 being not at all burdensome and 100 being very burdensome. Participants rated how much they agreed that the app was time consuming, taxing, and tedious (response options: strongly disagree, disagree, neutral, agree, or strongly agree). We dichotomized responses as strongly agree/agree versus strongly disagree/disagree/neutral.
Weight was obtained at baseline and at 12 weeks via the Wi-Fi scales sent to participants upon enrollment.
Participant engagement is defined as participant posts, replies, reactions (eg, love, wow, angry, and sad), and participation in intervention polls, which are used either as a way of assessing participant knowledge (eg, pop quizzes) or as a way for participants to share their diet and/or exercise barriers. We extracted engagement data from the private Facebook group using the Grytics app, except poll data, which were manually extracted because Grytics does not capture poll data. We summarized the total number of original posts, replies, reactions, and polls that each participant participated in. In addition, we calculated the percentage of participants in each condition who replied to each of the 12 weekly weigh-in posts.
We summarized retention, app use, slips, usability, acceptability, burden, and engagement in the Facebook groups, including the percentage participating in the weekly weigh-ins using descriptive statistics. For variables that were normally distributed, we described distributions using mean and SD, and for variables that were not normally distributed, we described distributions using median and IQR. We compared use, retention, usability, acceptability, and engagement by treatment condition using
A total of 846 individuals initiated the eligibility screening survey (
Consolidated Standards of Reporting Trials diagram. FB: Facebook.
Characteristics of participants by treatment condition (N=64).
Characteristics | Treatment condition | ||
|
Slip Buddy (n=32) | Calorie tracking (n=32) | |
Age (years), mean (SD) | 39.5 (9.6) | 40.2 (12.3) | |
BMI at enrollment (kg/m2), mean (SD) | 34.9 (5.3) | 33.4 (4.4) | |
Female, n (%) | 24 (75) | 24 (75) | |
|
|||
|
Non-Hispanic White | 26 (81) | 26 (81) |
|
Hispanic or Latino (any race) | 1 (3) | 4 (13) |
|
Non-Hispanic Black | 3 (9) | 1 (3) |
|
Non-Hispanic Asian | N/Aa | 1 (3) |
|
Other race or multiracial | 2 (6) | N/A |
|
|||
|
Married or living with a partner | 24 (75) | 22 (69) |
|
Single | 8 (25) | 7 (22) |
|
Divorced or separated | N/A | 3 (9) |
|
|||
|
At most high school | 2 (6) | N/A |
|
Trade or technical school, some college, or associate degree | 8 (25) | 10 (31) |
|
Bachelor’s degree | 9 (28) | 4 (13) |
|
Some graduate coursework | 4 (13) | 7 (22) |
|
Graduate degree | 9 (28) | 11 (34) |
|
|||
|
Employed full time | 22 (69) | 25 (78) |
|
Employed part time | 8 (25) | 5 (16) |
|
Homemaker (not looking for a job) | 2 (6) | N/A |
|
Student | 1 (3) | 4 (13) |
|
Retired | N/A | 1 (3) |
aN/A: not applicable.
bParticipants could select more than 1 employment status. In the Slip Buddy condition, 1 participant was employed part time and a student. In the calorie tracking condition, n=1 was employed part time and a student, n=1 was employed full time and a student, and n=1 was retired and employed part time.
Retention was high in both treatment conditions, with 97% (31/32) of Slip Buddy participants and 94% (30/32) of calorie tracking participants providing follow-up data (
Nearly all participants randomized to the Slip Buddy condition (31/32, 97%) and the calorie tracking condition (31/32, 97%) used their assigned app at least once during the 12-week intervention. Participants in the Slip Buddy condition used their assigned app on a mean of 44.0 (SD 25.8) days. Participants in the calorie tracking condition used their app on a mean of 46.3 days (SD 24.0), which represented use on an average of 53.8% (SD 31.3%) of possible days for participants in the Slip Buddy condition and 57.5% (SD 28.4%) of possible days for participants in the calorie tracking condition (t62=0.495;
App use by Slip Buddy participants was nearly identical in a sensitivity analysis assessing app use by self-report data when available, and when self-report data were not available, backend data were used. Overall, 97% (30/31) of participants used the app at least once during the 12-week intervention. They used the app on an average of 43.8 (SD 25.7) days, representing 53.4% (SE 31.2%) of possible days; 55% (17/31) of the participants used the app at least once in week 12 of the intervention.
One participant did not use the Slip Buddy app, and 1 participant responded to the notification but did not record any slips. The remaining participants reported a total of 538 slips during the 12-week intervention period. Participants who reported slips (n=30) reported a median of 15 slips (IQR 8-23; range 2-66; mean 17.9, SD 14.4). Most of slips happened at home (297/538, 55.2%), followed by work (113/538, 21.0%) and restaurant/bar (86/538, 16.0%;
Location, activity, eating episode, stress, and satiety associated with slips reported by participants over 12 weeks (N=538 slips).
Slip characteristics | Value, n (%) | ||
|
|||
|
Home | 297 (55.2) | |
|
Work | 113 (21) | |
|
Restaurant or bar | 86 (16.0) | |
|
Another person’s house | 22 (4.1) | |
|
The car | 10 (1.9) | |
|
An event | 8 (1.5) | |
|
The gym | 2 (0.4) | |
|
|||
|
Work or studying | 153 (28.4) | |
|
Socializing | 130 (24.2) | |
|
Screen time | 123 (22.9) | |
|
Domestic activities | 112 (20.8) | |
|
Commuting | 20 (3.7) | |
|
|||
|
Snack | 220 (40.9) | |
|
Dinner | 102 (19.0) | |
|
Dessert | 77 (14.3) | |
|
Lunch | 60 (11.2) | |
|
Breakfast | 40 (7.4) | |
|
Alcohol | 39 (7.3) | |
|
|||
|
Lower range, 0-4 | 298 (55.4) | |
|
Higher range, 5-10 | 240 (44.6) | |
|
|||
|
Hungry range, 0-4 | 299 (55.6) | |
|
Full range, 5-10 | 239 (44.4) |
The mean SUS score for the Slip Buddy condition was 64.8 (SD 16.5), which is the marginally acceptable range. Comparatively, the mean SUS score for participants in the calorie tracking condition was 76.3 (SD 17.6), which is considered good acceptability. Participants in the calorie tracking condition rated the MyFitnessPal app as more usable on average than the Slip Buddy participants rated the Slip Buddy app (t59=2.64;
Among participants who completed the follow-up survey, 39% (12/31) of Slip Buddy participants agreed or strongly agreed that tracking slips was helpful, whereas 77% (23/30) of calorie tracking participants agreed or strongly agreed that tracking diet and exercise was helpful (
Acceptability of assigned tracking app by treatment condition.
Acceptability itema | Slip Buddy (n=31), n (%) | Calorie tracking (n=30), n (%) | ||
Tracking my slips with Slip Buddy app/tracking my diet and exercise with MyFitnessPal was helpful for me | 12 (39) | 23 (77) | 8.9 (1) | .003 |
Tracking diet slips on Slip Buddy app/MyFitnessPal is easy | 24 (77) | 26 (87) | 0.9 (1) | .35 |
Using the Slip Buddy app/MyFitnessPal is tedious | 9 (29) | 12 (40) | 0.8 (1) | .37 |
Using the Slip Buddy app/MyFitnessPal is taxing | 5 (16) | 8 (28) | 1.2 (1) | .28 |
Using the Slip Buddy app/MyFitnessPal is time consuming | 6 (20) | 13 (43) | 3.8 (1) | .05 |
aProportion of participants responding with
Most of both Slip Buddy (24/31, 77%) and calorie tracking participants (26/30, 87%) agreed or strongly agreed that using their respective app was easy (
On a scale of 0 to 100, the median burden rating for participants in the Slip Buddy condition was 30 (IQR 15-50), and the median burden rating for participants in the calorie tracking condition was 45 (IQR 10-60; Mann-Whitney
A total of 88% (28/32) of participants in the Slip Buddy condition attended postintervention focus groups, and they made a total of 35 responses to the question about what they liked most about the Slip Buddy app (IRR=94%; κ=0.89). The most common theme of responses was the ease of use/simple concept (23/28, 66%), followed by increasing accountability and/or awareness of overeating and/or triggers (8/28, 23%), the end-of-day reminder to track slips (2/28, 5%), feeling motivated to not slip so there would be nothing to track (1/28, 3%), and other (1/28, 3%). Participants made a total of 35 responses about what they liked least about the Slip Buddy app (IRR=91%; κ=0.89). Responses reflected the following themes: technical issues (eg, app crashing and notifications not going away; 10/35, 29%), easy to forget to use or not sure how to use when no slips (7/35, 21%), did not find relevant stress ratings (5/35, 15%), focus on slips was too negative (4/35, 12%), was not sure what to count as a slip (4/35, 12%), did not include diet instruction (4/35, 12%), and slip history screen was not as informative as it could be (1/35, 3%). For the final question regarding their thoughts on a feature that would allow them to track when they were tempted but did not slip, 75% (21/28) said they would be enthusiastic about this feature. The remainder said they worried that it would add too much burden.
Over 12 weeks, participants randomized to the Slip Buddy condition had an average weight loss of −6.5 lb (SD 9.7) or 3.0% (SD 4.5%) of their baseline weight, and participants randomized to the calorie tracking condition had a weight loss of −7.5 lb (SD 10.7) or 3.6% (SD 4.9%) of their baseline weight (
Weight change from baseline to 12 weeks, by treatment condition.
Weight variablesa | Slip Buddy (n=32) | Calorie tracking (n=32) |
Baseline weight (lb)b, mean (SD) | 217.2 (39.0) | 208.5 (37.2) |
Follow-up weight (lb), mean (SD) | 210.7 (39.5) | 201.0 (36.3) |
Absolute weight change (lb), mean (SD) | −6.5 (9.7) | −7.5 (10.7) |
Percentage weight change, mean (SD) | −3.0 (4.5) | −3.6 (4.9) |
5% or greater weight loss, n (%) | 10 (31) | 11 (34) |
3% or greater weight loss, n (%) | 15 (47) | 15 (47) |
aWe used the last available weight from the study scales for 8% (5/64) participants. In Slip Buddy condition, 3% (1/32) participant became pregnant (used week 3 weight), and 3% (1/32) participant did not provide follow-up weight (week 6 weight). In the calorie tracking condition, 3% (1/32) participant became pregnant (we used week 2 weight), and 6% (2/32) participants did not provide follow-up weight (we used weeks 9 and 10 weight).
b1 lb = 0.45 kg.
In the Slip Buddy condition, the median total replies per participant was 55.00 (IQR 11.75-79.00), which was not significantly different from the median total replies of 29.5 (IQR 17.25-60.75;
In terms of weekly weigh-in participation, on average, 58.07% (SD 14.32; range 34%-81%) of Slip Buddy participants and 51.30% (SD 18.48; range 25%-88%) of calorie tracking participants replied to weigh-in posts each week (
Findings revealed that although participants in both treatment conditions used their assigned apps on a similar percentage of intervention days (ie, 54% of days for participants in the Slip Buddy condition and 58% of days for participants in the calorie tracking condition), 55% (17/31) of Slip Buddy participants used the app at week 12 of the intervention compared with only 35% (11/31) of calorie tracking participants. However, this difference was not statistically significant. Less than one-third of Slip Buddy participants agreed that using the Slip Buddy app was tedious (9/31, 29%), taxing (5/31, 16%), or time consuming (6/31, 20%); 77% (24/31) agreed that tracking slips was easy; but only 39% (12/31) agreed that tracking their slips with the app was helpful, which was significantly lower than that in the calorie tracking app condition. Slip Buddy also received lower usability ratings than the commercial calorie tracking app, perhaps not surprisingly, as commercial apps are years ahead of Slip Buddy in user experience optimization. Slip Buddy participants reported barriers such as technical difficulties, forgetting to use the app if they had not slipped in a while, and finding the exclusive focus on slips to be too negative. Slip data revealed that most of the slips reported happened at home, followed by work, and that snacks and dinner time were the eating episodes at which slips were most likely to occur. Activities that co-occurred with slips were distributed fairly evenly across work, socializing, screen time, and domestic tasks. Less than half of the slips occurred under conditions of moderate to high stress and over half occurred while hungry.
The Slip Buddy app was designed to reduce dietary self-monitoring to possibly its simplest form by only necessitating the recording of aberrant eating episodes. Despite its simplicity, use rates in this study appeared fairly comparable with app use in the commercial calorie tracking app condition. Interestingly, a randomized trial that compared a commercial calorie tracking app (Calorie Counter by Fat Secret) with the lower-burden Meal Logger app, which allows users to track their diet by taking photos of what they eat [
Our focus group data revealed that the infrequency in which the Slip Buddy app needed to be used (ie, only when a slip occurs) may have led some people to forget to use it. Slip Buddy participants, on average, only recorded an average of 1 to 2 lapses per week, which is far less than daily and possibly insufficient to capture enough calories to lose weight even if all slips were successfully eliminated; however, without calorie data, this is unknown. Studies using ecological momentary assessment to have people track dietary lapses have reported means ranging from 2.7 to 11.8 lapses per week [
Most participants wanted the Slip Buddy app to include the ability to track instances when they resisted temptations to give them an opportunity to see improvement in their ability to deal with contextual factors that cue slips. Including temptation tracking might allow users to develop a more regular habit of monitoring their eating habits and the circumstances in which they make healthy and unhealthy choices. In another study, participants in a behavioral weight loss program were asked to record their temptations and lapses in a paper diary in the final week of the program. They found that temptations were more likely than lapses to be followed by coping behaviors, suggesting the value of having people record both [
Despite the vast literature on emotional eating [
This pilot feasibility trial was not powered to detect group differences in weight loss, the primary outcome planned for the larger fully powered randomized trial. Statistical comparisons of clinical outcomes in pilot feasibility trials are also inappropriate because of the inflated risk of type 1 and type 2 errors [
This study had several limitations. First, the sample size was not large enough to compare the 2 conditions for weight loss. However, the purpose of this work was to evaluate the feasibility and acceptability of the Slip Buddy app using a user-informed process to guide improvements to the technology before conducting a fully powered randomized trial. We chose a 12-week intervention length to allow us to gain user insights after a prolonged period of use, but this study does not provide information on tracking habits over more extended periods, such as whether slip tracking decreases over time as participants learn their triggers and slip less. Another limitation is that our ability to measure app use via backend data was hampered by the fact that some participants did not see or receive end-of-day notifications that, when clicked, would indicate whether no slip entries meant a slip-free day or nonuse. For this reason, we had to rely on self-report use data that are prone to biases because of forgetting or social desirability. This can be ameliorated in the next version, and the addition of temptation and exercise slip tracking will give the user more to do each day with the app, even in the absence of slips. Another limitation is that the Slip Buddy app was not operating for 2 days, which could have affected use in subsequent days to the extent that participants were frustrated by this. As it is a newly developed app, bugs and crashes are more common than commercial apps that have been around for many years, and this will certainly impact usability ratings. An additional limitation is that both Slip Buddy and MyFitnessPal allow users to track other behaviors that may impact weight (eg, exercise, sleep, and mood), and we know little about the use of these features and whether their use impacted outcomes. Finally, a limitation is that the sample overrepresented non-Hispanic White women (40/64, 63% of our sample). In future research, recruitment will need to limit enrollment of non-Hispanic White women, the population segment that too often comprises most of the weight loss trial samples [
Although human counseling is associated with better outcomes in technology-delivered behavioral weight loss interventions [
Proportion of participants using their assigned app each week of the 12-week intervention, by treatment condition.
CONSORT-EHEALTH checklist (V 1.6.1).
Diabetes Prevention Program
interrater reliability
System Usability Scale
Weight Watchers
This work was funded by National Institutes of Health grants R01HL122302 and K24HL124366 awarded to SP.
SP is a paid scientific advisor for Fitbit and has been a paid consultant for WW.