The mLIFE randomized trial examining the impact of gamifying social support provision for weight loss

Abstract

Objective

The Mobile Lifestyle Intervention for Food and Exercise (mLIFE) study was a 12‐month mobile weight loss intervention examining social gaming to promote social support.

Methods

Adults with overweight or obesity (n = 243) were randomized to the mLIFE + points (received points for social support activities) or mLIFE group (blinded to points). Weight was measured via Fitbit scales. Repeated‐measures mixed models were used to conduct both an intent‐to‐treat analysis and an analysis among adherent participants (logged on to the mLIFE app ≥25% of study days).

Results

Attrition was lower in the mLIFE + points group (22% vs. 41% in mLIFE; χ² = 9.8, p < 0.01), and adherence was higher (61% vs. 42% in mLIFE; χ² = 7.6, p < 0.01). None of the group × time interactions was significant for the intent‐to‐treat analysis except for total number of points earned at 12 months (mLIFE + points mean 605.9 [SE 203.2] vs. mLIFE mean 350.0 [SE 200.0]; p < 0.01). The mLIFE + points participants lost a mean of 5.3 [SE 0.6] kg at 12 months (vs. mean 3.5 [SE 0.7] kg in mLIFE; p = 0.09). Among adherent participants (n = 127), mLIFE + points participants lost more weight (mean 7.3 [SE 0.8] kg) than mLIFE (mean 3.8 [SE 0.9] kg; p < 0.01).

Conclusions

The use of points led to greater weight loss at 12 months, but only among adherent participants. Providing points for completing social support activities is a way to retain participants and increase engagement in a mobile intervention.

Adults with overweight or obesity (n = 243) were randomized to the mLIFE + points (points for social support activities) or mLIFE group (no points). mLIFE + points participants lost 5.3 ± 0.6 kg at 12 months (vs. 3.5 ± 0.7 kg mLIFE; p = 0.09). Among adherent participants (n = 127), mLIFE + points participants lost more weight (7.3 ± 0.8 kg) than mLIFE (3.8 ± 0.9 kg; p < 0.01).

Study Importance.

What is already known?

• Provision and receipt of social support have been shown to be important components in both weight loss and diabetes‐related interventions.

• It has been challenging, however, to find ways to facilitate social support when translating behavioral weight loss programs to a mobile health (mHealth) delivery.

What does this study add?

• The use of points for provision of social support led to greater weight loss at 12 months, but only among adherent participants.

How might these results change the direction of research or the focus of clinical practice?

• Points for completing social support activities is a way to retain participants and increase engagement in a mobile intervention.

• Future programs for people with overweight or obesity may want to include the use of points as a way to engage and retain patients in clinical practice.

INTRODUCTION

Overweight and obesity are linked to a number of chronic diseases, including type 2 diabetes mellitus (T2DM) [1]. In traditional group‐based behavioral treatment of weight loss, several core components have proven to be important for success, including diet, physical activity (PA), and weight self‐monitoring [2] and group meetings [3]. In addition, social support plays an important role, which “provides a combination of empathy, social support, and healthy competition” [4]. Social support‐based weight loss interventions have shown to be effective for reducing body weight in interventions lasting longer than 6 months [5]. It has been challenging, however, to find ways to facilitate social support when translating behavioral weight loss programs to a mobile health (mHealth) delivery. Finding ways to encourage participants to provide social support to one another may be one strategy to ensure social support provision is occurring in a weight loss intervention.

mHealth delivery of behavioral weight loss interventions has numerous advantages, such as being cost‐effective and highly scalable [6]. An mHealth delivery format also allows for the integration of novel approaches to sustain motivation and engagement. These approaches include social gaming, which uses game‐like elements in nongame contexts to promote supportive social interactions and influence adoption of healthy behaviors [7]. Although gamified behavioral interventions have demonstrated improvements in body weight and hemoglobin A1c [8], there has been little research that has isolated the effect of gamifying social support [9].

The Mobile Lifestyle Intervention for Food and Exercise (mLIFE) study was a 1‐year randomized parallel behavioral weight loss intervention, delivered entirely via the mLIFE smartphone app. The mLIFE app focused on ways to promote social support for self‐monitoring diet, weight, and PA. The gamified intervention tested whether providing participants with points for completing social support activities led to greater weight loss than a group that did not receive points. The hypothesis for the main study outcome is that the group receiving points (mLIFE + points) would lose more weight at 12 months than the non‐gamified group (mLIFE). Secondary outcomes for the mLIFE study included blood pressure, waist and hip circumference, energy intake (kilocalories), PA, and points earned. The hypothesis for the secondary outcomes is that the mLIFE + points group would have greater improvements in these outcomes and earn more points at 12 months than the mLIFE group.

METHODS

The mLIFE study occurred between 2022 and 2024 and was entirely remotely delivered. The methods and a full description of the intervention for the mLIFE study, including inclusion and exclusion criteria, along with recruitment methods, have been described elsewhere [10]. Briefly, the mLIFE study was a 1‐year two‐arm randomized parallel behavioral intervention. Participants had to be between the ages of 18 and 65 years, live in the United States, be at risk for developing T2DM (must have overweight or obesity [body mass index, i.e., BMI, between 25 and 49.9 kg/m²] and two additional risk factors for T2DM [11]), own a smartphone, be willing to reduce energy intake and increase PA, not be taking weight loss medications, and be free of major health or psychiatric diseases or drug or alcohol dependency. Participants were recruited via online ads and via https://www.researchmatch.org. Participants were directed to a study website where they could complete an online screening form. Participants who qualified were then called to complete a phone screener and, if they qualified, were invited to attend a virtual live orientation session where they could learn more about the intervention content and assessment measures and complete a consent form. Once the consent form was completed, participants began assessment measures. All research was approved by the University of South Carolina Institutional Review Board, and mLIFE was overseen by a data safety officer.

Measures

All measures were assessed remotely and occurred at three time points: baseline, 6 months, and 12 months [10]. Participants received an Amazon gift card at both the 6‐ and 12‐month assessment time points to incentivize completion of all study measures. A baseline survey collected demographics. Changes in medications were assessed at each time point via Research Electronic Data Capture (REDCap; https://projectredcap.org/). Energy intake was assessed via three unannounced 24‐h dietary recalls using the Automated Self‐Administered 24‐h (ASA24) Dietary Assessment Tool (https://epi.grants.cancer.gov/asa24/) at each time point.

The main outcome of body weight was assessed via the provided Fitbit Aria Air scale (Google). Readings from the scale synced to the mLIFE app database via the participant's Bluetooth phone connection. Participants were instructed to weigh themselves upon waking without clothing and before eating or drinking anything but after voiding their bladder and bowels. Previous research has demonstrated that the use of e‐scales is highly correlated with clinic‐measured weight [12]. Occasionally, participants' phones would not sync with their scale. In these cases, participants could enter their weight into the Fitbit app and, at assessment time points, were asked to provide a photo of the weight on their scale as a verification of this measurement. PA was assessed via the provided Fitbit Versa 2 (Google), as well as the Exercise Vital Sign Questionnaire, which provided minutes of PA per week and has demonstrated validity in measuring PA as compared with accelerometry [13]. For the Fitbit PA data, which were pulled via the Fitbit application programming interface, a valid day of wear was considered if the participant wore the device for at least 10 h/day for at least 3 days during the week of the assessment [14]. For day‐level wear time tracking, Fitbit aggregates the hourly wear time data and reports the total wear time throughout the day [14]. Output from the devices used for analysis included steps per day and minutes of moderate‐to‐vigorous PA (MVPA) per week [15]. Previous studies have found Fitbit devices to be a valid way to measure PA, particularly steps, but may overestimate intensity [16].

Methods for assessing other outcomes have been described elsewhere [10]. Participants were asked to use an electronic blood pressure machine (either at home, in a physician's office, or at a pharmacy) to take their blood pressure reading after a 5‐min rest. Participants were instructed to take two readings and take a photo of each reading that could be uploaded to the REDCap survey. Participants also received a stretch‐resistant tape measure in the mail and viewed training videos on how to measure their own waist and hip circumferences.

Intervention data were collected from the mLIFE app. The number of points obtained throughout the study was assessed in both groups (points were only visible to the mLIFE + points group). In addition, minimum adherence to the study was defined as logging onto the app at least 25% of the days during the study. A log‐on was defined as opening the app and, at a minimum, viewing the home screen. Definitions of adherence to mobile behavioral interventions have varied widely [17]. Logging on ≥25% days was chosen to have sufficient power and to represent a minimal intervention needed for change threshold [18].

Intervention

Participants were randomized to one of two groups in blocks of 10 (with blocks created using a computerized random‐number generator by a study statistician) and stratified by sex and race (White vs. non‐White) to ensure equal distribution of both men and women and White and non‐White participants within groups. Once all baseline data were collected, a blinded study coordinator assigned participants to the next number in the block. Participants did not learn of their randomized assignment until they attended their first training session, in which they learned about the mLIFE app.

Both groups downloaded the mLIFE app, in which they tracked their dietary intake and received intervention content. In addition, all participants received a Fitbit Versa 2 watch to track PA and a FitBit Aria scale to track weight. PA and weight values from the Fitbit devices were displayed on graphs within the mLIFE app for participants to view. The app contained a news feed that showed participants' activities (e.g., “Lisa just logged her weight for the day”). Participants could provide a thumbs up for participants who completed their activities. The behavioral content of the intervention was delivered within the mLIFE app through podcasts (two per week) and tips of the day (five per week) and have been described in detail elsewhere [10]. The content of both the podcasts and tips of the day was informed by the Diabetes Prevention Program [19] and the Social Cognitive Theory and has been used in several previous studies [10].

The two groups differed in one way: The mLIFE + points group received points for social support activities and could see those points whereas the mLIFE participants could do the same activities but could not see their points. (However, the app tracked the points that would have been earned so researchers could view them.) Specifically, the mLIFE + points participants had a leaderboard in which they could track their points and see how they related to other members of their group. The mLIFE + points group received up to eight points per day for completing social support activities (1. “liking” a fellow participant's tracking of diet, 2. exercise, or 3. weight or 4. when a fellow participant listened to a podcast or read a tip of the day; 5. sending an encouraging message to a participant who had not logged onto the app in 24 h; 6. sending or 7. responding to a request for support related to diet or exercise; and 8. thanking a participant for support received). The mLIFE group did not have a leaderboard. The leaderboard reset each month so participants in the mLIFE + points group started the competition over.

In order to allow staff time to recruit and start a group of participants at the same time, the mLIFE study was conducted in two cohorts, separated by 1 year, and the intervention was similar in both cohorts except for diet tracking. The mLIFE app had a built‐in diet tracking interface that was used for cohort 1. However, it did not contain some of the features of diet tracking that some participants desired, such as bar code scanning and saving recipes. For cohort 2, participants tracked their diet in the Fitbit app, and the dietary data then appeared within the mLIFE app.

Statistical analysis

The study was powered on the main outcome of weight loss and was calculated to achieve 95% power (α = 0.05) assuming a small effect size (d = 0.1) and an intracluster correlation equal to 0.7, which led to a sample size equal to 200. To account for 20% attrition, the study aimed to recruit 240 participants. Complete power calculations have been presented elsewhere [10].

The statisticians for the study were blinded to group assignment when completing all analyses. Descriptive statistics were used to present baseline characteristics. Because body weight was collected via three potential methods (a synced scale reading, manual entry with photo validation, or, in cases in whcih neither a phone sync nor photo was available, manual entry without photo validation), one‐way ANOVA and χ² test were used to examine differences in weight loss among participants’ method of reporting body weight at 6 and 12 months and differences in method between groups to ensure weight loss by reporting method and method of tracking weight did not differ between groups. Intent‐to‐treat (ITT) analysis was conducted, and the primary and secondary outcomes were addressed using repeated measures. Mixed models with maximum likelihood estimation and robust computation of standard errors (SE) using the SAS system version 9.4 (SAS Institute Inc.) provide estimates interpretable under the ITT principle. The same models were conducted using multiple imputation with related outcome variables through Markov chain Monte Carlo simulation and did not lead to different results. Therefore, the results using the repeated‐measures models are presented.

The distributional assumption for outcomes was checked. The models included time, group, and a group × time interaction, and full information from available data was used. Contrasts were constructed comparing body weight (or other outcomes) at 6 and 12 months between groups and to estimate the changes from baseline to each time point. The model included covariates adjusting for baseline education (≥college graduate or <college graduate), sex, age, and time‐specific medication use (for blood pressure outcomes). Although the groups were stratified by sex, sex was still included in the models due to evidence that men find goal setting less important than women [20] and adhere more to self‐monitoring [21]. The same analyses were conducted among a subset of participants who were at least minimally adherent to the intervention (logged on at least 25% of study days). Effect sizes (d) were calculated for changes between group and within group in each outcome.

RESULTS

A total of 243 adults with overweight or obesity (mean BMI 34.7 [SE 6.0] and mean age 46.7 [SE 11.6] years, 87% female, 67% White/22% Black) enrolled in the mLIFE study (Figure 1), and baseline demographics are presented in Table 1. All participants are included in the ITT analysis, regardless of whether they provided follow‐up data. A weight measure (primary outcome) was available for 199 participants at 6 months (18% attrition) and 164 participants at 12 months (33% attrition). A total of 137 participants (56%) successfully synced their weight at 6 months, 52 participants (21%) entered their weight manually and verified entry with a photo, and 10 participants (4%) entered their weight manually but did not upload a photo. At 12 months, 120 participants (49%) synced their weight, 24 participants (10%) manually entered with a photo verification, and 20 (8%) manually entered a weight without uploading a photo. Weight loss did not differ among the three methods at either 6 (F = 1.62, p = 0.20) or 12 months (F = 1.69, p = 0.17), and method of weight entry did not differ between groups at either 6 (χ² = 2.64, p = 0.27) or 12 months (χ² = 0.31, p = 0.86). Therefore, participants who reported their weight with any method were included in the analysis. Attrition at 12 months was lower in the mLIFE + points group (22% vs. 41% mLIFE group; χ² = 9.8, p < 0.01). Participants were adherent on a mean of 40.9% [SE 35.9%] of days and a median of 27% (interquartile range [IQR] 0.27%–100%), and more participants were adherent in the mLIFE + points group (mean 61%, median 36%) than the mLIFE group (mean 42%, median 19%; χ² = 7.6, p < 0.01). One participant (mLIFE group) was excluded from analysis due to developing a thyroid disorder (which was an exclusion criterion).

FIGURE 1.

CONSORT (Consolidated Standards of Reporting Trials) flow diagram for the mLIFE (Mobile Lifestyle Intervention for Food and Exercise) randomized trial. [Color figure can be viewed at wileyonlinelibrary.com]

TABLE 1.

Baseline characteristics of study participants in the mLIFE study.

	mLIFE + points group	mLIFE group
n	121	122
Age, mean ± SE, y	47.0 ± 1.0	46.4 ± 1.0
Sex, n (%)
Female	107 (88.4)	110 (90.2)
Male	14 (11.6)	2 (9.8)
Race, n (%)
American Indian/Alaska Native	0 (0)	2 (1.5)
Asian	5 (4.0)	5 (4.0)
Native Hawaiian or other Pacific Islander	0 (0)	0 (0)
Black or African American	29 (24.0)	26 (21.0)
White	81 (67.0)	85 (69.6)
More than one race	6 (5.0)	3 (2.4)
Unknown or not reported	0 (0)	2 (1.5)
Ethnicity, n (%)
Hispanic or Latino	11 (9.0)	11 (9.0)
Not Hispanic or Latino	110 (91.0)	110 (90.2)
Unknown	0 (0)	1 (0.8)
Education, n (%)
High school or equivalent or some college	24 (19.8)	36 (29.5)
College graduate	48 (39.7)	40 (32.8)
Advanced degree	49 (40.5)	46 (37.7)
Occupation, n (%)
Employed for wages	90 (74.4)	90 (73.7)
Contract/temporary	4 (3.3)	3 (2.5)
Retired	12 (9.9)	11 (9.0)
Other a	15 (12.4)	18 (14.8)
Marital status, n (%)
Single	25 (20.7)	23 (18.8)
Married	68 (56.2)	79 (64.8)
Divorced or separated	15 (12.4)	10 (8.2)
Widowed	3 (2.4)	4 (3.3)
Other b	10 (8.3)	6 (4.9)
BMI, mean ± SE, kg/m2	34.9 ± 0.5	34.6 ± 0.5
Body weight, mean ± SE, kg	106.8 ± 2.4	104.4 ± 2.4
Systolic blood pressure, mean ± SE, mm Hg	128.2 ± 1.8	130.0 ± 1.8
Diastolic blood pressure, mean ± SE, mm Hg	81.8 ± 1.2	81.7 ± 1.3
Waist circumference, mean ± SE, cm	111.7 ± 4.8	114.2 ± 4.9
Hip circumference, mean ± SE, cm	119.2 ± 2.4	116.9 ± 2.5
Energy intake, mean ± SE, kcal/d	2136.2 ± 79.6	2073.9 ± 80.1
Steps/d via Fitbit, mean ± SE	6868.4 ± 375.8	7038.3 ± 386.7
MVPA via Fitbit, mean ± SE, min/wk	261.12 ± 45.10	220.27 ± 43.19
PA from Exercise Vital Sign Questionnaire, mean ± SE, min/wk	179.3 ± 28.6	128.1 ± 30.1

Abbreviations: mLIFE, Mobile Lifestyle Intervention for Food and Exercise; MVPA, moderate‐to‐vigorous physical activity; PA, physical activity.

^a Other for employment includes being unable to work or seeking opportunities or prefer not to say.

^b Other for marital status includes living with someone.

None of the group × time interactions was significant for the ITT analysis (least squares mean [SE]) except for total number of points earned at 12 months (Table 2). Participants could earn ≤2920 points across the 12‐month study. The mLIFE + points participants earned more points for engaging in social support activities over the course of the study (605.9 [203.2]) compared with mLIFE participants (350.0 [200.0]; p < 0.01). Differences in weight loss were not significant, with mLIFE + points participants losing 4.9 [0.6] kg at 6 months (vs. 3.6 [0.7] kg in mLIFE; p = 0.15) and 5.3 [0.6] kg at 12 months (vs. 3.5 [0.7] kg in mLIFE; p = 0.09). At 6 month, 39% of participants in the mLIFE + points group achieved ≥5% weight loss (28% in mLIFE), and at 12 months, 35% of mLIFE + points participants achieved 5% weight loss (24% in mLIFE group). Results in both groups went in the hypothesized direction apart from waist circumference and PA. The mLIFE + points group had a significant within‐group increase in waist circumference at 6 months, and the mLIFE group has a significant within‐group increase at 12 months. All PA measures decreased at both 6 and 12 months, with significant within‐group decreases in Fitbit‐measured MVPA and steps at 6 months in the mLIFE group and significant 6‐month within‐group decreases in self‐reported PA for the mLIFE + points group. At 6 months, 43% of participants in the mLIFE + points group achieved a mean of ≥150 min/week of MVPA (24% in mLIFE), and at 12 months, 41% of mLIFE + points participants achieved a mean of ≥150 min/week of MVPA (28% in mLIFE).

TABLE 2.

Changes in weight, blood pressure, body circumference, energy intake, and PA and number of points earned between the mLIFE + points and the mLIFE groups at 6 and 12 months, presented as adjusted least squares mean ± SE (95% CI), effect size (d) ^a

	mLIFE + points group (n = 121)	mLIFE group (n = 122)	Difference between groups	Effect size for difference between groups (d)	p value for difference between groups
Weight change, kg
6 mo	−4.87 ± 0.61 (−6.07 to −3.66)* d = −0.24	−3.56 ± 0.66 (−4.85 to −2.26)* d = −0.18	−1.31 ± 0.9 (−3.08 to 0.46)	−0.21	0.15
12 mo	−5.30 ± 0.64 (−6.56 to −4.05)* d = −0.26	−3.49 ± 0.72 (−4.92 to −2.07)* d = −0.19	−1.81 ± 0.97 (−3.71 to 0.09)	−0.28	0.09
Change in systolic blood pressure, mm Hg
6 mo	−3.23 ± 1.51 (−6.20 to −0.26)* d = −0.24	−4.01 ± 1.65 (−7.27 to −0.76)* d = −0.24	0.78 ± 2.24 (−3.62 to 5.19)	0.04	0.73
12 mo	−2.79 ± 1.51 (−5.76 to 0.19) d = −0.20	−5.00 ± 1.64 (−8.23 to −1.78)* d = −0.30	2.21 ± 2.23 (−2.17 to 6.61)	0.15	0.32
Change in diastolic blood pressure, mm Hg
6 mo	−0.51 ± 1.21 (−2.88 to 1.86) d = −0.05	−1.94 ± 1.31 (−4.52 to 0.65) d = −0.18	1.43 ± 1.78 (−2.08 to 4.94)	0.13	0.42
12 mo	−2.75 ± 1.21 (−5.12 to −0.37)* d = −0.29	−2.81 ± 1.3 (−5.37 to −0.24)* d = −0.26	0.06 ± 1.78 (−3.44 to 3.56)	0.00	0.97
Change in waist circumference, cm
6 mo	13.84 ± 4.69 (4.62 to 23.07)* d = 0.48	4.71 ± 5.11 (−5.34 to 14.77) d = 0.14	9.13 ± 6.93 (−4.52 to 22.77)	0.26	0.19
12 mo	9.08 ± 4.71 (−0.18 to 18.35) d = 0.31	10.43 ± 4.98 (0.63 to 20.24)* d = 0.30	−1.35 ± 6.85 (−14.84 to 12.14)	−0.03	0.84
Change in hip circumference, cm
6 mo	−4.57 ± 2.7 (−9.88 to 0.75) d = −0.29	−0.45 ± 2.92 (−6.2 to 5.3) d = −0.03	−4.11 ± 3.98 (−11.94 to 3.72)	−0.11	0.30
12 mo	−4.06 ± 2.71 (−9.39 to 1.27) d = −0.26	−1.03 ± 2.87 (−6.67 to 4.61) d = −0.07	−3.03 ± 3.94 (−10.79 to 4.74)	−0.13	0.44
Change in energy intake, mean kcal/d
6 mo	−241.93 ± 85.47 (−409.98 to −73.87)* d = −0.35	−361.39 ± 88.98 (−536.34 to −186.43)* d = −0.50	119.46 ± 123.39 (−123.15 to 362.07)	0.21	0.33
12 mo	−347.73 ± 91.15 (−526.96 to −168.5)* d = −0.50	−257.42 ± 92.93 (−440.14 to −74.69)* d = −0.36	−90.31 ± 130.21 (−346.33 to 165.70)	−0.09	0.49
Change in steps, mean steps/d via Fitbit
6 mo	−32.79 ± 263.7 (−551.42 to 485.85) d = −0.01	−690.13 ± 275.27 (−1231.52 to −148.73)* d = −0.25	657.34 ± 381.17 (−92.35 to 1407.03)	0.24	0.09
12 mo	441.58 ± 288.14 (−125.14 to 1008.31) d = 0.14	102.58 ± 314.3 (−515.59 to 720.75) d = 0.04	339 ± 425.91 (−498.68 to 1176.68)	0.12	0.43
Change in MVPA, mean min/wk via Fitbit
6 mo	−31.62 ± 21.72 (−74.36 to 11.11) d = −0.15	−78.90 ± 22.47 (−123.12 to −34.69)* d = −0.45	47.28 ± 31.25 (−14.21 to 108.78)	0.20	0.13
12 mo	−4.66 ± 23.45 (−50.81 to 41.48) d = −0.02	−3.50 ± 24.91 (−52.52 to 45.53) d = −0.02	−1.17 ± 34.18 (−68.43 to 66.10)	−0.01	0.97
Change in PA, min/wk from Exercise Vital Sign Questionnaire
6 mo	−67.07 ± 26.28 (−118.73 to −15.4)* d = −0.21	−15.77 ± 27.98 (−70.78 to 39.23) d = −0.07	−51.3 ± 38.38 (−126.76 to 24.17)	−0.21	0.18
12 mo	−29.09 ± 26.82 (−81.82 to 23.64) d = −0.09	−19.71 ± 28.25 (−75.25 to 35.84) d = −0.08	−9.38 ± 38.95 (−85.96 to 67.2)	−0.04	0.81
Total points earned
6 mo	283.28 ± 203.22 (−117.16 to 683.73)	146.4 ± 199.96 (−247.63 to 540.43)	136.88 ± 81.54 (−23.79 to 297.55)	0.41	0.09
12 mo	605.94 ± 203.22 (205.49 to 1006.38)	349.96 ± 199.96 (−44.07 to 744)	255.97 ± 81.54 (95.3 to 416.64)	0.40	0.002

Abbreviations: mLIFE, Mobile Lifestyle Intervention for Food and Exercise; MVPA, moderate‐to‐vigorous physical activity; PA, physical activity.

^a All models adjusted for baseline sex, age, and education and use of medications that may impact the examined outcome.

^*Significant within‐group changes, p < 0.05.

For the analysis among adherent participants (n = 127), 6‐month weight loss did not differ (6.1 [0.8] kg in mLIFE + points vs. 4.1 [0.9] kg in mLIFE; p = 0.09) but weight loss at 12 months did. Adherent mLIFE + points participants lost significantly more weight (7.3 [0.8] kg) than adherent mLIFE participants (3.8 [0.9] kg; p < 0.01). None of the other outcomes for adherent participants was significant and they were in similar directions to the outcomes for all participants (Table S1).

DISCUSSION

The 12‐month mLIFE intervention examined the use of points to gamify provision of social support during a remotely delivered behavioral weight loss intervention, and points led to higher engagement with the study and less attrition. Although the outcomes did not differ when analyzing the entire sample, when examining participants who used the app for at least 25% of intervention days, there was a difference in weight loss such that mLIFE + points participants lost 3.5 kg more than mLIFE participants. Because the use of points reduced attrition, it may be important to find additional ways to use points in mHealth weight loss interventions (such as rewarding streaks for logging diet and weight or providing badges when a participant hits a certain point level). In addition, assessing potential motivators of engagement at baseline (such as assessing participant preferences for different types of engagement approaches) could allow for a more tailored approach to gamifying an intervention.

The lack of a difference in weight loss in the ITT analysis may be due to a control comparison that was very similar to the intervention group. In a meta‐analysis of mobile weight loss interventions, the mean difference between mHealth conditions and conditions with minimal intervention was 2.5 kg at 12 months [22] (difference in the present study was 1.8 kg). However, the 12‐month weight loss seen in both groups is higher than what has been observed in other mHealth weight loss interventions [23]. This study highlights that rewarding points to participants who provide social support led to better adherence to the intervention and, among those who were adherent, better weight loss.

Much of the previous research around social support in weight loss has focused on in‐person interventions [24, 25]. As more behavioral weight loss treatment programs move to digital platforms, finding ways to provide and encourage social support is important [26]. A previous mHealth study found that higher levels of informational or emotional support provided by users were associated with higher completion of diet and PA self‐monitoring by other users [27]. In addition, although self‐monitoring diet, PA, and weight has been shown to promote weight loss [21], an innovative aspect of mLIFE is that participants were rewarded for encouraging and congratulating others on completing these behaviors, adding both aspects of social support and encouragement to self‐monitoring. Participants were also incentivized to interact with one another to improve engagement, which has shown to be effective in promoting weight loss [28]. Provision of points may be one way to promote greater engagement and social support and less attrition in mHealth weight loss interventions.

Although gamification is commonly found in technology for education [29] or in commercial entertainment apps, use of gaming elements in mHealth is still a growing area [30]. Gamified interventions for weight loss have typically gamified completion of self‐monitoring and weight goals [31, 32]. A unique feature of the mLIFE study was that points were not rewarded for completing behaviors but for acknowledging and supporting others who completed health behaviors.

One of the primary ways behavioral interventions use gamification is through the use of points [9]. The mLIFE study empirically tested whether using points to promote social support led to greater weight loss than the same intervention without points. Results suggest the provision of points for completing social support activities was an important way to engage and retain participants. By the end of the study, mLIFE + points participants had accrued 288 more points than the mLIFE group (which couldn't see any points earned). Because the mLIFE + points group had less attrition and, among adherent participants only, more weight loss, provision of points for completing social support activities is a potential way to retain participants and increase engagement in a mobile intervention.

The mLIFE study used a gain‐framed approach, in which participants received points for behaviors, versus a loss‐framed approach, in which participants might lose points for not completing behaviors. Gain‐framed approaches highlight the benefits of a given health behavior, whereas loss‐framed approaches highlight the disadvantages of a behavior [33, 34]. Previous research on using gain versus loss framing in nutrition, PA, and weight loss studies has been mixed. Some studies have shown that gain‐framed messaging is more effective than loss‐framed messaging in the area of diet and nutrition for both preventing disease [34] and encouraging adherence [35]. However, other studies have found that using a loss‐framed approach may be effective for improving PA [31]. Future studies examining the role of providing social support during a mobile weight loss intervention might want to compare a gain‐ versus loss‐framed approach.

Although not significant between groups, some of the outcomes worsened over time. For example, waist circumference increased in both groups. Studies examining the use of self‐assessed waist circumference have found strong correlations with clinically measured waist circumference, but participants tend to overestimate self‐assessed measures by a mean of 6 cm [36]. Waist circumference can increase despite weight loss as evidenced in the Action for Health in Diabetes (Look AHEAD) behavioral weight loss trial, which found that 10% of participants had increases in waist circumference while still losing weight [37]. In addition, both objective and self‐reported PA decreased, especially from baseline to 6 months. Six‐month PA was assessed in both cohorts between Thanksgiving and Christmas holidays. Studies have found that PA tends to decline around the winter holidays [38]. Although self‐report methods of PA tend to lead to overreporting of PA [39], in the present study, objective measures of PA mirrored the self‐reported PA directions.

The mLIFE study has several strengths. The study recruited a nationwide sample, increasing the generalizability of the findings and allowing for a diverse sample (23% Black and 9% Latino/Hispanic). The design of the intervention and app allowed for an empirical test of the use of points to incentivize provision of social support. Objective measures of weight and PA were collected using Fitbit devices, and three 24‐h dietary recalls using multiple pass methodology were used at each time point, which is considered an optimal way to assess energy intake [40].

The mLIFE study also has several limitations, which should be taken into consideration when interpreting the findings. Despite a nationwide recruitment, the participant sample ended up being mostly highly educated, White, female participants. In a meta‐analysis of 12 mHealth weight loss interventions, 11 of the studies included mostly women [41], and in a review of technology‐based interventions that included at least 50% racial and ethnic minority adults, only 6 studies were identified [42]. Future studies may want to employ additional recruitment measures to attract men and African American participants. Although the use of Fitbits provided an objective and valid way to track PA [16], the algorithms used to calculate MVPA are proprietary [43]. However, mLIFE was interested in detecting differences between groups, and the Fitbits calculated PA levels similarly across all participants. In addition, although participants were trained to collect both blood pressure and body circumference measures on their own, having a trained assessor who was blinded to condition would have been a more reliable way to obtain these measures. Lastly, the mLIFE study had higher attrition (32%) than was predicted (20%), which may have led to it being underpowered to detect differences. For example, effect sizes for both between‐ and within‐group weight loss were <0.3, and differences between the groups at 12 months were p = 0.09. However, attrition rates can be high in mHealth‐delivered interventions, with reviews finding attrition rates between 20% [44] and 40% [45]. Lastly, the study did not take into account potential influences of outside social support, such as spouses, friends, or family members.

CONCLUSION

The mLIFE study was a rigorous test of gamifying a weight loss intervention by awarding points to participants for providing social support. Although no differences in outcomes were found, there was higher engagement (as measured through points earned) and better retention among the mLIFE + points group. When examining the outcomes among a subsample of participants who logged on to the app at least 25% of the study days, mLIFE + points participants lost 3.5 kg more than the non‐points mLIFE group. Because provision of points is an easy way to gamify a weight loss intervention, future behavioral treatment programs may wish to use points to enhance retention, engagement, and weight loss. In addition, future research should examine the relationship between weight loss and objectively measured perceptions of social support, along with social support provision and receipt.

AUTHOR CONTRIBUTIONS

All coauthors conceived the project. Data acquisition, analysis, and interpretation were conducted by Gabrielle M. Turner‐McGrievy, Diana Carolina Delgado‐Díaz, Kelli E. DuBois, Yesil Kim, and James Hardin. Gabrielle M. Turner‐McGrievy, Diana Carolina Delgado‐Díaz, Kelli E. DuBois, Halide Zeynep Aydin, Courtney M. Monroe, Homayoun Valafar, and Sara Wilcox were involved in the design and implementation of the intervention. Yesil Kim and James Hardin performed statistical analyses. All authors contributed substantial portions of writing to the manuscript. Gabrielle M. Turner‐McGrievy obtained funding for the study. All authors provided critical revision of the manuscript and approved the final version.

FUNDING INFORMATION

The research reported in this publication was supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under award number R01DK129302. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

CONFLICT OF INTEREST STATEMENT

The authors declared no conflicts of interest.

CLINICAL TRIAL REGISTRATION

ClinicalTrials.gov identifier NCT05176847.

Supporting information

Table S1. Changes in weight, blood pressure, body circumference, energy intake, and physical activity, and number of points earned between adherent^a mLife + points participants and adherent mLife participants at six and 12 months presented as adjusted least squares mean±standard error (95% CI), effect size (d)^b.

Turner‐McGrievy GM, Delgado‐Díaz DC, DuBois KE, et al. The mLIFE randomized trial examining the impact of gamifying social support provision for weight loss. Obesity (Silver Spring). 2025;33(8):1447‐1456. doi: 10.1002/oby.24330

DATA AVAILABILITY STATEMENT

Individual participant data that underlie the results reported in this article after deidentification (text, tables, figures, and appendices) will be made available upon request beginning 3 months and ending 5 years following any final publications from this study by the research team.