Abstract
Low-socioeconomic status (SES) individuals have higher rates of obesity. Social media platforms are used frequently by low-SES individuals and facilitate important weight loss program components including social support. Very few social media-based weight loss interventions, however, have enrolled or been tailored to low-SES participants. The purpose of this article is to examine the feasibility of a social media-based weight loss intervention among low-SES adults. We conducted a one-group pretest post-test pilot intervention study with two groups (group 1, n = 39, group 2, n = 16) of low-SES overweight/obese adults who were enrolled in a 12-week social media-based weight loss intervention including self-monitoring via Fitbits and participation in a private Facebook group. A moderator provided educational content and encouraged social support via Facebook. Descriptive statistics were used to assess intervention acceptability and engagement. Exploratory analyses were conducted to examine changes in study outcomes and engagement patterns. The study had good retention (86%). Among 55 total participants enrolled, there were 9,175 participant interactions within the Facebook group. Among completers (n = 47), 96% indicated they would recommend the intervention to a friend. Mean weight loss was 1.07 kg (SD = 3.96, p = .0498), and participants reported increases in positive dietary social support (mean = 2.47, SD = 5.09, p = .0007). Engagement in this social media-based pilot intervention was high and exceeded results from previous studies using similar formats. Participants evaluated the intervention favorably. Changes in weight loss and several theoretical mediators were in the desired direction. Overall, our results indicate social media groups as a platform for weight loss intervention delivery among low-SES adults are feasible and should be studied in larger randomized trials.
Keywords: Weight loss, Social media, Health disparities, Dissemination
Adults with low income found a weight loss program that used Facebook valuable, contributed a large amount of content to the Facebook weight loss group, and reported increased social support.
Implications.
Practice: Social media-based interventions can be used to engage low-socioeconomic status participants in theoretically relevant activities that support weight loss.
Policy: Social media-based weight loss intervention approaches that are low cost and easily disseminated could have positive implications for program delivery to at-risk populations, but efficacy needs to be established before policy action is considered.
Research: Future research should be aimed at identifying the effectiveness of social media-based interventions to enhance weight loss among low-socioeconomic status groups.
Introduction
Individuals with low socioeconomic status (SES) are more likely to be obese and suffer disproportionately from several chronic diseases including cancer, cardiovascular disease, and diabetes [1, 2]. Despite this, low-SES groups have been largely underrepresented in weight loss intervention studies [3–5]. This highlights the need for the development of effective weight loss strategies for low-SES individuals, which could reduce chronic disease risk in a particularly vulnerable population.
A potentially effective way of delivering weight loss interventions to low-SES participants is the use of social media platforms (e.g., Facebook, Twitter). Social media-based interventions offer advantages inherent in web-based delivery including continuous availability and remote access, which could reduce barriers to intervention use among low-SES participants such as limited time or transportation [6, 7]. Over half of U.S. internet users with high school or less education (64%) and income less than $30,000 per year (68%) use social media [8]. Seventy-one percent of U.S. individuals within these demographic categories also report owning a smartphone [9], which is the primary device by which U.S. users access social media [10].
Social media tools create novel opportunities for delivering evidence-based behavior change strategies. Communication features of social media websites encourage users to contribute and react to each other’s content. Furthermore, survey data suggest social media users are accustomed to sharing information about their health experiences [11]. By facilitating the sharing of health information, such as weight loss goals and experiences as well as dietary and physical activity behavior, social media interventions are well suited to the delivery of important behavior change techniques including social support and behavioral modeling [12].
Preliminary evidence from behavioral trials suggests social media-based interventions can be efficacious in producing weight loss, increase the frequency of social interaction versus traditional approaches, and positively affect theoretical mediators such as social support [13]. Although the number of social media-based weight loss interventions enrolling low-SES participants are limited [14], several recent studies have enrolled low-income women in weight loss interventions using social media components. Two of those studies reported a reduction in excessive gestational weight gain and greater postpartum weight loss among intervention participants [15, 16]. In a series of pilot studies conducted by Silfee et al., postpartum women were enrolled in three Facebook-delivered DPP-based behavioral weight loss intervention. Weight loss across the pilot studies ranged from 2.6 to 7.0 lbs over 16 weeks and high levels of participant satisfaction were reported [17]. Examination of recruitment between two similar DPP-based interventions, one delivered via Facebook and one delivered in a face-to-face format, also found that low-SES participants were equally receptive to Facebook and in-person delivery [18].
Based on the limited existing literature examining the use of social media-based weight loss interventions among low-SES adult populations, the importance of establishing effective weight loss interventions for low-SES adults, and the potential of using social media for such a purpose, we conducted a 12-week pilot intervention study to assess the feasibility of a weight loss program delivered via social media to low-SES adults. Primary measures were intervention acceptability and patterns of engagement. Exploratory, secondary analyses were also conducted to investigate changes in outcomes, relationships between changes in outcomes and engagement, and associations between engagement and intervention content.
MATERIALS AND METHODS
Design
This study employed a single-group pretest–post-test design enrolling two separate groups. Study procedures were the same for each group except where noted. All procedures were approved by the Case Western Reserve University Institutional Review Board.
Participants
Recruitment
Participants were recruited between October 2017 and February 2018 and received the intervention in two separate groups (group 1, January–March, 2018; group 2, March–May 2018). We recruited via advertisements on social media platforms including Facebook, Twitter, and Instagram. Seven advertisements were created that included demographically diverse images and were targeted by age group and geographic location. In addition, flyers were placed in strategic locations (e.g., public libraries) throughout the recruitment area and emails were sent to a university employee listserv. We also recruited in-person at Supplement Nutrition Assistance Program Education (SNAPED) and Expanded Food and Nutrition Education Program (EFNEP) events. Participants were considered eligible if they met the following inclusion criteria: aged 35–65 years, regular internet access as defined by having access to an internet enabled device at least 2 hr per day on four or more days per week, body mass index (BMI) between 25 and 40, less than 30 min per day of moderate or vigorous physical activity, not pregnant or planning a pregnancy, not currently being treated for schizophrenia, not currently or in the past 2 years being treated for alcohol or drug abuse, live in the target metropolitan area in Northeast Ohio as defined by zip code, income less than 185% of federal poverty based on family size, and did not answer yes to any questions on the Physical Activity Readiness Questionnaire (PAR-Q) [19] or received written clearance by a physician. This age group was chosen based on the assumption that individuals in a more limited age group would have greater homophily, which is associated with both greater support seeking and perceptions of higher support quality in online groups [20, 21]. Because our study included an unsupervised exercise component, minimal contact with participants, and no planned medical supervision, we chose to exclude individuals categorized as morbidly obese (BMI ≥ 40), who face unique barriers to participating in traditional exercise protocols and increased health risk [22, 23].
Study procedures
Enrollment, orientation, and follow-up
Advertisements included a link to web-based screener questionnaire that assessed eligibility. Those who answered yes to any PAR-Q questions but who were otherwise eligible were emailed a physician clearance form and instructed to have their doctor complete it and send it to study staff. If deemed eligible, participants received a link via email to a second questionnaire. This questionnaire began with an online consent form. If an individual provided consent, they could continue and complete the baseline measures. Consented individuals were also invited to attend an orientation meeting. Orientation meetings were held at locations convenient to potential participants (e.g., public libraries). At the orientation meeting, we collected anthropometric data and determined final eligibility. If eligible, participants were enrolled in a Facebook group for use solely by study participants and provided with a digital scale and Fitbit Flex activity tracker. Participants also completed an individualized educational session where study staff provided information about exercise safety, Facebook group privacy and participation, use of the Fitbit activity tracker, and other study requirements and procedures. Participants received a $30 gift card as compensation for attending the orientation session. Upon completion of the 12-week intervention, participants were invited via email to complete a follow-up questionnaire and were invited to attend a follow-up meeting. At the follow-up meeting, anthropometric data were collected and a $50 gift card was provided to participants for completing the study.
INSHAPE CLE intervention
The 12-week INSHAPE CLE intervention was based on a previously tested face-to-face group-based behavioral weight loss intervention [24] that emphasized lifestyle behavior changes and modest caloric deficit. The intervention was delivered via a Facebook group and participant use of the Fitbit self-monitoring wearable device and goal setting and tracking features. The settings for this Facebook group prevented contributions to the group from being shared with a participant’s larger Facebook social network and allowed the moderator to only admit study participants into the group. This format was chosen to ensure the privacy of participants and encourage sharing of personal information related to weight loss. Intervention content developed a priori by the study team was delivered by messages posted to the Facebook group by a moderator. The moderator was a Master’s-level nutrition student who was trained by the Principal Investigator and instructed to post all planned, previously developed communications, monitor the group for harmful content, and respond to participant questions during the intervention in consultation with the Principal Investigator. This overall approach was chosen so as to encourage participants to contribute the majority of support, thereby examining the proposition that similar interventions could be automated in future iterations [25]. Intervention content used several modalities, including text, photos, links, and videos. Planned moderator content also used a variety of engagement strategies: Education Only—posts that only contained educational information, Recipes—posts with recipes that asked participants to contribute their own recipes, Testimonials/Goal Setting—posts that asked participants to share goals and program progress and setbacks, “Ask a Dietitian”—posts that encouraged participants to ask questions about diet and exercise, Content Announcements—previews of upcoming program topics, Competition—posts that announce winners from group and individual competitions, Education + Question—posts that have educational content but also ask participants questions, Question Only—posts that ask participants questions without educational content, and Memes. Planned posts were delivered approximately three times per day during the intervention. For group 1, the moderator posted 283 times and commented 144 times, for group 2, moderator contributions were 268 and 24, respectively. Two hundred fifty-four posts were the same between groups. The discrepancy between groups is a result of different questions asked by participants and some changes implemented in the communications related to challenges for group 2. For group 1, a limited number of likes were contributed by the moderator (n = 51) on participant posts and comments to encourage group participation in the beginning days of the intervention. Given the robust, spontaneous response from participants, this was not implemented for group 2.
Measures
Demographics
As a part of the baseline questionnaire, participants were asked to report their birth year, gender, race, ethnicity, educational status, and employment status.
Recruitment and retention
Recruitment and retention were assessed by recording the number of participants screened and their recruitment origin, eligible participants, participants who provided informed consent, participants who attended orientation, and participants who attended the follow-up meeting.
Intervention acceptability
Acceptability was measured using a 19-item questionnaire based on a previous instrument developed by the authors [26]. The questionnaire assessed participant attitudes toward intervention activities and study procedures including recruitment, enrollment, online questionnaires; the type and frequency of communications, and barriers to participation. For example, participants were asked how valuable a series of activities were to them during the intervention (e.g., competitions) with a 5-point scale anchored by “Extremely important” and “Not at all important.”
Intervention use
Data were collected on Facebook group use including the content of the posts and comments (e.g., “Next week we’re going to be talking about barriers to eating right and exercising enough. Take a minute to think about your own barriers to being healthy (we all have them). What is the biggest challenge you face?”), a time stamp, and any reactions (e.g., emojis that include likes, loves, hahas, sads, angries, and wows) to or comments on posts or comments. Posts, comments, and reactions were summed at the individual level to produce a variable representing engagement (interactions). Participants were split into two groups (<1 interaction per day, low engagement and ≥1 interaction per day, high engagement) to create a categorical variable for engagement. Two data collection methods were used to obtain Facebook data. The Grytics analytical platform (https://grytics.com/) provided data directly from the Facebook API. We also collected select data manually by reviewing and logging information directly from the Facebook group. This was required due to changes to Facebook’s privacy policy that occurred during the second group’s intervention that de-identified Facebook user data. In addition, the Grytics platform did not provide data on reactions for comments with the exception of likes. Data were validated by comparing aggregate data provided by Grytics, where available, to the totals from the manual data collection. In cases where totals did not agree, Facebook was manually checked a second time. Overall, 48% of reactions were compared. Of those (2,880), Grytics underreported 57 reactions (2%). There were no discrepancies in the total number of posts or comments between Grytics data and manual data. Reaction data from two Facebook posts were deleted from Facebook before manual collection occurred. Aggregate reaction data for those posts are included in post-level data analysis but not individual level analysis. Data on the use of the Fitbit for self-monitoring were collected via the Fitabase research platform (http://www.fitabase.com). We calculated two variables from the Fitabase data, the number of days that participants reported any steps and the number of times participants logged their dietary intake during the intervention.
Anthropometrics
All anthropometrics were collected by study staff who attended a 1-hr training session. We measured height using a stadiometer (Detecto PHR Portable Mechanical Height Rod for DR400C) and weight using a floor scale (Detecto DR550C) calibrated using a 10kg weight (Ohaus 80850302). Measurement procedures were based on the NHANES Anthropometry Procedures Manual [27]. All measurements were taken twice and a third measurement was taken if the difference between the first two exceeded 0.3 cm for height and 0.1 kg for weight. Once three measurements were taken, the two closest were averaged to determine the measurement.
Other measures
Dietary knowledge was assessed by the sum of correct answers from a 36-item questionnaire adapted from a validated nutrition knowledge questionnaire [28] and was scored as the percentage answered correctly. The questionnaire assessed several domains of nutrition knowledge including national dietary recommendations, the nutrient content of foods, and the relationship between dietary behavior and chronic disease. Social support was measured by the 5-item Friend Social Support and Eating Habits Scale [29] that asks participants to indicate how often their friends (1 = never; 5 = very often) communicated positive messages about dietary behavior (e.g., “Discussed my eating habit changes with me (asked me how I’m doing with my eating changes)”). We used a validated four-item scale measure to assess weight loss self-efficacy [30]. This scale asked participants “How confident are you that you can lose weight?” with a series of conditional statements (e.g., “Even if you have to try several times until it works”) rated on a sliding scale from (0% = not at all confident; 100% = completely confident).
Data analysis
Data analyses were completed using SPSS v24 (IBM Corp, Armonk, NY). Data were screened for deviations from assumptions required for the statistical analyses used. We calculated descriptive statistics for baseline characteristics, enrollment and retention, acceptability, and outcome measures. Baseline differences between completers and those lost to follow up as well as differences between study groups were examined using t-tests for continuous variables and chi-square tests for categorical variables. We performed paired-samples t-tests to explore changes in outcome and mediator variables and one-sample t-tests to generate confidence intervals for percent weight loss. We examined the association between participant baseline characteristics (BMI, race, education, and age) and participant interactions as well as associations between moderator post characteristics (post modalities and engagement strategies) and participant interactions using negative binomial regression, which is used for analyzing over dispersed count outcome variables [31]. One-way ANOVA was used to explore associations between participant engagement (high vs. low) and changes in study outcomes and mediators. We also examined correlations between the number of different types of engagement (posts, comments, and reactions) and study outcomes and mediators. Primary analyses were conducted on the entire sample (n = 55) using a baseline observation carried forward approach for examining changes in study outcomes and mediators. We also examined results for study completers (n = 47) as a secondary, less conservative analysis.
RESULTS
Participants
Figure 1 shows recruitment and retention. Of 2,340 individuals screened, 1,844 (79%) were identified as being recruited via social media; 130 (6%) were eligible. The most common reasons for ineligibility were not meeting SES criteria (67%), answering yes to any PARQ question without physician clearance (43%), and having BMI > 40 kg/m2 (43%). Of those enrolled in the study (N = 55), 86% (n = 47) were retained at follow-up. Participants enrolled in the Facebook group (n = 55) were predominantly female (95%) and majority African American (69%) with a mean age of 47 years. Most participants reported being employed (66%) and almost half reported completing college or graduate school (42%) (see Table 1 for details). There were no significant differences in demographic characteristics between completers and those lost to follow-up. Completers in group 1 were less likely to be unemployed than completers in groups 2, χ 2 (5, N = 47) = 16.35, p = .0059. No other differences were observed between groups.
Fig 1.
Recruitment and retention of low-socioeconomic status adults in a feasibility pilot study of a Facebook-delivered weight loss intervention.
Table 1.
Participant demographics for INSHAPE CLE participants
| Enrolled in Facebook group | Completers | |||||
|---|---|---|---|---|---|---|
| Group 1 (n = 39) | Group 2 (n = 16) | Total (n = 55) | Group 1 (n = 34) | Group 2 (n = 13) | Total (n = 47) | |
| Age, mean (SD) | 45.95 (8.69) | 48.06 (10.97) | 46.56 (9.35) | 45.59 (8.91) | 48.38 (10.52) | 46.36 (9.35) |
| Gender, N (%) | ||||||
| Female | 36 (92.3) | 16 (100.0) | 52 (94.5) | 31 (91.2) | 13 (100.0) | 44 (93.6) |
| Male | 2 (5.1) | 0 (0) | 2 (3.6) | 2 (5.9) | 0 (0) | 2 (4.3) |
| Transgender | 1 (2.6) | 0 (0) | 1 (1.8) | 1 (2.9) | 0 (0) | 1 (2.1) |
| Hispanic, N (%) | ||||||
| Yes | 2 (5.1) | 1 (6.3) | 3 (5.5) | 1 (2.9) | 1 (7.7) | 2 (4.3) |
| No | 32 (82.1) | 15 (93.8) | 47 (85.5) | 28 (82.4) | 12 (92.3) | 40 (85.1) |
| Did not know/would rather not say | 3 (7.7) | 0 (0) | 3 (5.5) | 3 (8.8) | 0 (0) | 3 (6.4) |
| Did not answer | 2 (5.1) | 0 (0) | 2 (3.6) | 0 (0) | 0 (0) | 2 (4.3) |
| Race, N (%) | ||||||
| White | 10 (25.6) | 3 (18.8) | 13 (23.6) | 9 (26.5) | 3 (23.1) | 12 (25.5) |
| Black or African American | 26 (66.7) | 12 (75.0) | 38 (69.1) | 23 (67.6) | 9 (69.2) | 32 (68.1) |
| More than one race | 3 (7.7) | 1 (6.3) | 4 (7.3) | 2 (5.9) | 1 (7.7) | 3 (6.4) |
| Employment, N (%) | ||||||
| Disabled | 3 (7.7) | 0 (0) | 3 (5.5) | 3 (8.8) | 0 (0) | 3 (6.4) |
| Employed | 29 (74.4) | 7 (43.8) | 36 (65.5) | 26 (76.5) | 4 (30.8) | 30 (63.8) |
| Homemaker | 0 (0) | 2 (12.5) | 2 (3.6) | 0 (0) | 2 (15.4) | 2 (4.3) |
| Retired | 1 (2.6) | 2 (12.5) | 3 (5.5) | 1 (2.9) | 2 (15.4) | 3 (6.4) |
| Student | 1 (2.6) | 1 (6.3) | 2 (3.6) | 0 (0) | 1 (7.7) | 1 (2.1) |
| Unemployed | 5 (12.8) | 4 (25.0) | 9 (16.4) | 4 (11.8) | 4 (30.8) | 8 (17.0) |
| Education, N (%) | ||||||
| College graduate or more | 20 (51.3) | 3 (18.8) | 23 (41.8) | 19 (55.9) | 3 (23.1) | 22 (46.8) |
| Some college | 17 (43.6) | 9 (56.3) | 26 (47.3) | 13 (38.2) | 7 (53.8) | 20 (42.6) |
| High school graduate | 2 (5.1) | 4 (25.0) | 6 (10.9) | 2 (5.9) | 3 (23.1) | 5 (10.6) |
| BMI, mean (SD) | 34.02 (3.83) | 35.09 (3.53) | 34.33 (3.74) | 33.79 (3.88) | 34.87 (3.86) | 34.09 (3.86) |
| Overweight, N (%) | 9 (23.1) | 2 (12.5) | 11 (20.0) | 9 (26.5) | 2 (15.4) | 11 (23.4) |
| Obese, N (%) | 30 (76.9) | 14 (87.5) | 44 (80.0) | 25 (73.5) | 11 (84.6) | 36 (76.6) |
BMI body mass index.
Intervention use and acceptability
Table 2 details Facebook engagement and Fitbit use by group. Participants enrolled in a Facebook group (n = 55) interacted a total of 9,175 times with a median of 104 interactions (interquartile range [IQR] = 28–273) during the intervention. Participant contributions appeared to vary across day of the week, times of day, and over the course of the intervention (see Fig. 2). There was a median of 577.50 (IQR = 346.8–996) interactions per week over the course of the intervention and engagement decreased throughout the intervention until it rebounded at the end. Participants contributed more in the beginning part of the week and during the middle of the day. Participant Fitbit use appeared to be more consistent for logging steps (median of 79 days logged, IQR = 53–83) than logging weight (median of three times logged, IQR = 1–9).
Table 2.
Total engagement and engagement per INSHAPE CLE participant enrolled in a Facebook group (n = 55)
| Enrolled in Facebook group | |||
|---|---|---|---|
| Group 1 (n = 39) | Group 2 (n = 16) | Total (n = 55) | |
| Posts | |||
| Total number | 391 | 81 | 472 |
| Median | 5 | 1 | 4 |
| Interquartile range | 2–14 | 0–4.5 | 1–11 |
| Minimum | 0 | 0 | 0 |
| Maximum | 73 | 47 | 73 |
| Comments | |||
| Total number | 2,051 | 681 | 2,732 |
| Median | 32 | 26 | 31 |
| Interquartile range | 11–74 | 2.8–61.3 | 11–66 |
| Minimum | 0 | 0 | 0 |
| Max | 264 | 222 | 264 |
| Reactions | |||
| Total number | 4,503 | 1,468 | 5,971 |
| Median | 70 | 67 | 70 |
| Interquartile range | 15–187 | 5–178.8 | 15–187 |
| Minimum | 0 | 0 | 0 |
| Maximum | 463 | 312 | 463 |
| Interactions | |||
| Total number | 6,945 | 2,230 | 9,175 |
| Median | 104 | 92 | 104 |
| Interquartile range | 34–273 | 7.8–250.8 | 28–273 |
| Minimum | 4 | 0 | 0 |
| Maximum | 643 | 581 | 643 |
| Number of days steps logged | |||
| Total number | 2,526 | 1,138 | 3,664 |
| Median | 73 | 83.5 | 79 |
| Interquartile range | 51–83 | 75.3–84 | 53–83 |
| Minimum | 2 | 0 | 0 |
| Maximum | 83 | 85 | 85 |
| Number of times weight logged | |||
| Total number | 217 | 97 | 314 |
| Median | 4 | 3 | 3 |
| Interquartile range | 1–10 | 1–6.8 | 1–9 |
| Minimum | 0 | 0 | 0 |
| Maximum | 26 | 29 | 29 |
Fig 2.
Patterns of engagement among INSHAPE CLE participants enrolled in a Facebook group (n = 55).
Analysis of participant interactions by moderator post characteristics indicated significant differences in engagement (see Table 3). Engagement strategies used in planned posts by the moderator related to testimonials and goal setting (mean = 6.23, SD = 5.08) and education plus a question (mean = 5.52, SD = 5.43) received on average more comments from participants than education only (mean = 0.59, SD = 1.24). The same analysis in relation to producing reactions found that competition (mean = 5.84, SD = 3.51) and content announcements (mean = 5.29, SD = 3.67) produced more reactions on average than education only (mean = 4.17, SD = 2.21). When compared with text only content, both videos and photos produced on average more reactions (video; incidence rate ratio = 1.69, photo; incidence rate ratio = 1.50). No recorded demographic characteristics predicted participant interactions.
Table 3.
Associations between planned moderator post characteristics and INSHAPE CLE participant comments and reactions (n = 551)
| Total (%) | Comments | IRR | 95% CI | Reactions | IRR | 95% CI | |||
|---|---|---|---|---|---|---|---|---|---|
| Mean (SD) | Lower | Upper | Mean (SD) | Lower | Upper | ||||
| Modality | |||||||||
| Video | 44 (8.0) | 0.66 (0.99) | 0.16* | 0.09 | 0.31 | 4.27 (1.99) | 1.69* | 1.33 | 2.13 |
| Photo | 284 (51.5) | 3.17 (4.58) | 0.79 | 0.57 | 1.09 | 3.80 (2.85) | 1.50* | 1.30 | 1.74 |
| Link | 64 (11.6) | 0.97 (1.90) | 0.24* | 0.14 | 0.41 | 3.06 (1.80) | 1.21 | 0.97 | 1.51 |
| Text only (reference) | 159 (28.9) | 4.01 (5.93) | 1.00 | . | . | 2.53 (2.26) | 1.00 | ||
| Engagement strategy | |||||||||
| Meme | 24 (4.4) | 0.96 (1.30) | 1.62 | 0.81 | 3.24 | 4.38 (3.06) | 1.05 | 0.82 | 1.34 |
| Question only | 102 (18.5) | 4.66 (5.69) | 7.87* | 5.47 | 11.32 | 1.60 (1.27) | 0.38* | 0.32 | 0.46 |
| Education + question | 54 (9.8) | 5.52 (5.43) | 9.33* | 6.04 | 14.41 | 3.06 (1.97) | 0.73* | 0.60 | 0.89 |
| Competition | 25 (4.5) | 4.52 (7.47) | 7.64* | 4.27 | 13.67 | 5.84 (3.51) | 1.40* | 1.12 | 1.75 |
| Content announcement | 48 (8.7) | 1.54 (3.46) | 2.61* | 1.59 | 4.26 | 5.29 (3.67) | 1.27* | 1.07 | 1.51 |
| Ask a dietitian | 16 (2.9) | 4.69 (6.80) | 7.92* | 3.93 | 15.96 | 1.69 (1.40) | 0.41* | 0.27 | 0.61 |
| Testimonials/goal setting | 69 (12.5) | 6.23 (5.08) | 10.53* | 7.05 | 15.73 | 2.14 (1.79) | 0.52* | 0.42 | 0.63 |
| Recipes | 22 (4.0) | 1.23 (2.02) | 2.07* | 1.04 | 4.15 | 2.86 (1.39) | 0.69* | 0.51 | 0.92 |
| Education only (reference) | 191 (34.7) | 0.59 (1.24) | 1.00 | . | . | 4.17 (2.21) | 1.00 |
CI confidence interval; IRR incidence rate ratio.
*Significant based on 95% CI.
Completers rated the intervention favorably with 96% indicating that they would recommend the program to their friends. When asked to evaluate different Facebook components provided, study completers rated group polls (55%), recipes (62%), setting public goals (57%), testimonials (72%), and group challenges (62%) as extremely or very valuable. Participant also reported finding tracking their exercise (74%) and tracking their weight (68%) either extremely or very valuable.
Changes in study outcomes and mediators
Exploratory analysis of changes in outcomes found that weight loss among participants (n = 55) averaged 1.07 kg, p = .0498. Weight loss in group 1 (n = 39) was 1.21 kg versus 0.73 kg in group 2 (n = 16). Thirty-one (56.4%) participants lost weight and 9 (16%) lost ≥ 5% of their baseline weight (see Table 4 for details of changes in outcomes). One participant in group 1 gained 13.3 kg and was determined to be an outlier. Weight loss in group 1 (n = 38) was 1.59 kg (SD = 3.50); t(37) = −2.81, p = .0080, d = 0.46, and weight loss in groups 1 and 2 combined (n = 54) was 1.34 kg (SD = 3.47); t(53) = −2.84, p = .0064, d = 0.39 when the outlier was removed. Overall participants reported significantly increased positive social support for dietary change and increased dietary knowledge. Changes in self efficacy were negative but not statistically significant. Results from the secondary analysis of completers (n = 47) are included inTable 5. A comparison of mean change in study outcomes between participants split by those who engaged at least once per day versus those who did not (low vs. high engagers) found a statistically significant difference for weight loss self-efficacy (low mean = −7.75, SD = 19.19; high mean = 1.51, SD = 14.52); F(1, 53) = 4.12, p = .0474). All other comparisons were not significant, weight (low mean = −0.07, SD = 4.63; high mean = −1.97, SD = 3.06); F(1, 53) = 3.28, p = .0760), positive social support for dietary change (low mean = 1.35, SD = 4.19; high mean = 3.48, SD = 5.66); F(1, 53) = 2.48, p = .1209), and dietary knowledge (low mean = 1.31, SD = 5.89; high mean = 2.52, SD = 5.64); F(1, 53) = 0.61, p = .4401). A significant correlation was found between the number of reactions and changes in weight loss self-efficacy, r = .37, p = .0117. No other significant correlations were found between the number of different types of engagement (posts, comments, and reactions) and study outcomes.
Table 4.
Changes in study outcomes among INSHAPE CLE participants (n = 55)
| Outcome | n | Baseline | Follow-up | d | Mean change (95% CI) |
|---|---|---|---|---|---|
| Mean (SD) | Mean (SD) | ||||
| Group 1 | |||||
| Weight (kg) | 39 | 95.34 (12.71) | 94.12 (13.28) | −0.29 | −1.21 (−2.57, 0.15) |
| Weight change (%) | 39 | — | — | — | −1.26 (−2.56, 0.05) |
| Positive dietary social support (5–25) | 39 | 9.92 (5.22) | 12.74 (5.26) | 0.52 | 2.82 (1.06, 4.58)* |
| Weight loss self-efficacy (0–100) | 39 | 86.13 (15.77) | 83.51 (19.16) | −0.16 | −2.63 (−7.83, 2.58) |
| Dietary knowledge (0–100) | 39 | 28.64 (7.62) | 31.62 (7.27) | 0.54 | 2.97 (1.19, 4.76)* |
| Group 2 | |||||
| Weight (kg) | 16 | 95.49 (11.74) | 94.76 (13.46) | −0.21 | −0.73 (−2.55, 1.09) |
| Weight change (%) | 16 | — | — | — | −0.95 (−2.86, 0.97) |
| Positive dietary social support (5–25) | 16 | 10.88 (4.50) | 12.50 (3.69) | 0.39 | 1.63 (−0.59, 3.84) |
| Weight loss self-efficacy (0–100) | 16 | 89.39 (12.22) | 85.94 (20.86) | −0.17 | −3.45 (−14.53, 7.62) |
| Dietary knowledge (0–100) | 16 | 26.06 (6.07) | 25.50 (7.71) | −0.10 | −0.56 (−3.59, 2.46) |
| Total | |||||
| Weight (kg) | 55 | 95.38 (12.33) | 94.31 (13.21) | −0.27 | −1.07 (−2.14, 0.00)* |
| Weight change (%) | 55 | — | — | — | −1.17 (−2.22, −0.12)* |
| Positive dietary social support (5–25) | 55 | 10.20 (5.00) | 12.67 (4.82) | 0.49 | 2.47 (1.10, 3.85)* |
| Weight loss self-efficacy (0–100) | 55 | 87.08 (14.79) | 84.21 (19.50) | −0.17 | −2.87 (−7.56, 1.83) |
| Dietary knowledge (0–100) | 55 | 27.89 (7.24) | 29.84 (7.85) | 0.34 | 1.95 (0.39, 3.50)* |
CI confidence interval.
*Significant based on 95% CI.
Table 5.
Changes in study outcomes among INSHAPE CLE study completers (n = 47)
| Outcome | n | Baseline mean (SD) | Follow-up mean (SD) | d | Mean change (95% CI) |
|---|---|---|---|---|---|
| Group 1 | |||||
| Weight (kg) | 34 | 95.00 (13.27) | 93.61 (13.85) | −0.31 | −1.39 (−2.95, 0.17) |
| Weight change (%) | 34 | — | — | — | −1.44 (−2.94, 0.06) |
| Positive dietary social support (5–25) | 34 | 10.18 (5.37) | 13.29 (5.24) | 0.57 | 3.12 (1.12, 5.12)* |
| Weight loss self-efficacy (0–100) | 33 | 85.28 (16.35) | 82.68 (19.88) | 0.01 | −2.60 (−8.73, 3.53) |
| Dietary knowledge (0–100) | 34 | 28.24 (7.97) | 31.59 (7.63) | 0.58 | 3.35 (1.33, 5.38)* |
| Group 2 | |||||
| Weight (kg) | 13 | 93.70 (12.11) | 92.80 (14.01) | −0.24 | −0.90 (−3.19, 1.39) |
| Weight change (%) | 13 | — | — | — | −1.17 (−3.58, 1.24) |
| Positive dietary social support (5–25) | 12 | 11.33 (5.12) | 13.50 (3.71) | 0.46 | 2.17 (−0.84, 5.17) |
| Weight loss self-efficacy (0–100) | 13 | 88.17 (13.20) | 83.92 (22.73) | −0.18 | −4.25 (−18.25, 9.75) |
| Dietary knowledge (0–100) | 12 | 26.58 (6.44) | 25.83 (8.55) | −0.11 | −0.75 (−4.96, 3.46) |
| Total | |||||
| Weight (kg) | 47 | 94.64 (13.76) | 93.38 (13.76) | −0.29 | −1.25 (−2.51, 0.00)* |
| Weight change (%) | 47 | — | — | — | −1.37 (−2.59, −0.14)* |
| Positive dietary social support (5–25) | 46 | 10.48 (5.27) | 13.35 (4.85) | 0.53 | 2.87 (1.25, 4.49)* |
| Weight loss self-efficacy (0–100) | 46 | 86.10 (15.44) | 83.03 (20.48) | −0.16 | −3.07 (−8.67, 2.54) |
| Dietary knowledge (0–100) | 46 | 27.80 (7.57) | 30.09 (8.19) | 0.37 | 2.28 (0.44, 4.13)* |
CI confidence interval.
*Significant based on 95% CI.
Discussion
Our findings support the feasibility of using social media as a platform for delivering weight loss interventions to low-SES groups. Social media was an effective method for generating interest in the study, but the inclusion criteria dramatically reduced the number of participants enrolled. Participants actively engaged in the Facebook group through posts, comments, and reactions. In addition, participants rated their experience within the group favorably. Exploratory analysis indicated that changes in most mediators and outcomes were in the desired direction.
Difficulty recruiting and retaining low-SES study participants is a well-documented phenomenon in research studies [32]. Although we implemented strategies to improve the effectiveness of our recruitment including using several modes of recruitment, providing incentives, conducting recruitment and enrollment at geographically diverse community locations, and limiting the number of face to face visits required of participants, the pace of our recruitment affected intended study implementation. Social media recruitment did produce a large number of interested individuals, but the inclusion criteria may need to be relaxed where appropriate. Specifically, enrolling individuals with greater BMI and lower age and expanding the income requirements may increase enrollment while maintaining the purpose of aiding individuals with limited resources in need of effective weight loss strategies. Additionally, a study design that allows for more incremental recruitment, enlists community members, and increases the number of community organization may increase the pace of enrollment. Given the number of screened but ineligible participants, more precision in recruitment would lessen the number of rejected participants. However, all relevant social media advertisement targeting options were used to reach our intended population.
Participant engagement in the Facebook group used in this study exceeded previous studies using similar formats targeting diet and/or physical activity. We found a median of 104 interactions per Facebook enrollee during the intervention. By comparison, during a lifestyle behavior change intervention study conducted in 2016 that enrolled low-income women, participants contributed a mean of 8 times to the Facebook group over 20 weeks [33]. In a series of pilot weight loss studies conducted by Pagoto et al., participants contributed a median number of tweets in a dedicated weight loss Twitter group ranging from 66 to 74 over 12 weeks [34]. A recent pilot weight loss study enrolling postpartum women in a Facebook group reported a median of 2 posts, 24 replies, and 32 likes over 12 weeks [35]. A study solely targeting physical activity found a median of 2 Facebook interactions per young adult cancer survivor participant over 12 weeks [36]. The drop off in engagement over time detailed in this study is a common phenomenon that has been reported in a number of other social media-based weight loss studies [36, 37]. Despite that, in the current study engagement was still robust even in weeks with reduced participation and rebounded at the end of the intervention. Higher engagement levels in this study could be the result of several factors. The frequency of planned moderator contributions in this study were higher (approximately three planned communications per day) than other studies in which the number of planned daily communications ranged from approximately 1 to 2 per day [34–36, 38]. Social media use over the past 5 years has increased, especially among groups similar to those enrolled in the current study. According to 2019 Pew data, African American, low-income, and lower educations adults use of social media has increased 8%, 10%, and 14%, respectively, over that time period [8]. Greater use may lead to greater acceptability of and familiarity with the platform leading to greater engagement. In the current study, we recruited primarily through Facebook. Our sample may therefore be more predisposed to Facebook use. Few studies have examined engagement by day of the week or time of day, but our results suggest that interventionists should be attuned to variations in engagement related to these factors in intervention design and implementation. We also found that participants more frequently used reactions versus comments and that posts were the least used type of interaction with the Facebook group. Edney et al. reported a similar finding in that likes were the most common form of engagement found in their analysis of commercial organizations’ social media campaigns [39].
Engagement with social media intervention components has been positively associated with increased weight loss [40, 41], making the design of social media moderator messages an important design concern for social media-based interventions. Our findings suggest that different types of posts may enhance different forms of engagement. We found that videos and photos produced more reactions than links and text-only messages but that text-only messages produced the most comments. Previous studies examining these relationships have largely not separated forms of engagement but have found that the use of both photographs and videos in Facebook posts increases overall engagement [42–44]. Messages that solicited input from participants (e.g., those with direct questions or requests for information) produced more comments but that relationship was reversed for reactions. Other studies have also found that discussion questions and directly asking participants for feedback is more successful than other types of messages [36, 41].
Our findings raise the intriguing question of what types of engagement are most important. Traditionally, comments and posts have been considered more valuable than reactions in terms of engagement based on the increased effort they require (for instance the Grytics platform weights comments twice in their engagement score), but this is an important empirical question that is not well explored in the intervention literature. In the current study, we only found one significant correlation between changes in outcomes and types of reactions but no discernible pattern by reaction type. Furthermore, the practice of “lurking” or observing social media content but not interacting with it deserves similar consideration. Lurking is estimated to comprise 90% of social media behavior, representing a large portion of individual’s interactions with the platform [45].With the exception of weight loss self-efficacy, we did not observe a statistically significant association between participant engagement level and changes in study outcomes although the pattern of mean differences was in the hypothesized direction. This could be the result of limited statistical power or the lack of an association.
Weight loss in the current study was less than some other interventions targeting similar populations. In the Weight Wise study targeting toward low-income women, participants lost a mean of 3.7 kg over 5 months [46]. Another physician-based study with a similar focus produced a 2-kg weight loss over 6 months [47]. Our results could be attributed to any number of factors including shorter length of the intervention and/or lower intensity. We also had one outlying case of weight gain that had a large effect on mean weight loss. Moderator involvement with participants during the intervention was intentionally limited. Some evidence suggests that more frequent personal contact is associated with greater weight loss in technology-mediated interventions [48]. This raises the intriguing question of the relative effectiveness of engaging with the moderator versus other participants. Although this study produced greater participant engagement, it had lower moderator engagement. There were also differences in changes in weight and engagement between groups in this study, with group 1 seemingly experiencing better results. This could be the result of several factors. Group 1’s intervention started at the new year when many individuals have higher motivation to lose weight [49]. Also, the moderator’s initial contribution of “likes” for group 1 may have created different initial conditions for the group. Another possibility is that larger groups attain some critical mass that has a positive effect on engagement, which has been found when examining online communities [50]. Given the dissemination potential due to technology and the low variable cost for the intervention, however, we believe weight loss produced in this pilot study over 12 weeks is meaningful.
This study has several limitations. We used income as a proxy measure for SES. Given that 42% of our participants reported attaining a college degree or advanced degree, we cannot generalize our results to individuals with low levels of education. Similarly, we were unsuccessful in recruiting male participants, a situation common to behavioral weight loss interventions but that also limits the generalizability of our findings [51]. We did not conduct follow-up data collection with participants who were eligible but did not enroll, limiting our ability to understand barriers to enrollment. Our primary goal was to assess the acceptability and feasibility of delivering a weight loss intervention via social media to a group of low SES participants. For this reason, we chose a one group, pre-post design to maximize Facebook group enrollment. This design, however, weakens the conclusions we can draw about changes in weight and other outcomes due to the lack of a control group. In addition, the pace of recruitment required splitting our intervention into two groups, reducing the number of participants in a single group. Results from the examination of associations between engagement and changes in study outcomes could be the result of underlying motivation as opposed to a causal relationship. Our collection of Facebook engagement data was accomplished through two methods, and it is possible that errors may have occurred in the manual collection of data, although our comparison of manually collected data to aggregate server collected data increases our confidence in the consistency of these methods.
Results from this pilot study endorse further exploration of the use of social media platforms for weight loss interventions targeting low-SES adults using adequately powered randomized designs. Social media was particularly effective in engaging this population to contribute comments and reactions. Manual collection of “views” at the individual level by reviewing Facebook posts and comments is possible using the existing platform. Analysis of these data could begin to elucidate the influence of lurking behavior. Future studies should also examine the tradeoff between cost and dissemination and the intensity of moderator involvement in interventions that rely on participants to contribute intervention content and exchange support.
Dr. Webb Hooper is now the Deputy Director of the National Institute on Minority Health and Health Disparities (NIMHD),
Funding:
This study was funded by the National Cancer Institute via the Case Comprehensive Cancer Center (CON501837).
Compliance with Ethical Standards
Conflicts of Interest: M.W.H. is the Deputy Director of the National Institute on Minority Health and Health Disparities (NIMHD). D.N.C., R,M., and S.F.F. declare that they have no conflicts of interest.
Authors’ Contributions: All the authors have agreed with the contents of the disclosure report and have approved the final manuscript.
Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. All procedures were approved by the Case Western Reserve University Institutional Review Board. This article does not contain any studies with animals performed by any of the authors.
Informed Consent: Informed consent was obtained from all individual participants included in the study.
References
- 1. McLaren L. Socioeconomic status and obesity. Epidemiol Rev. 2007;29:29–48. [DOI] [PubMed] [Google Scholar]
- 2. Sommer I, Griebler U, Mahlknecht P, et al. Socioeconomic inequalities in non-communicable diseases and their risk factors: An overview of systematic reviews. BMC Public Health. 2015;15:914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Beauchamp A, Backholer K, Magliano D, Peeters A. The effect of obesity prevention interventions according to socioeconomic position: A systematic review. Obes Rev. 2014;15(7):541–554. [DOI] [PubMed] [Google Scholar]
- 4. McGill R, Anwar E, Orton L, et al. Are interventions to promote healthy eating equally effective for all? Systematic review of socioeconomic inequalities in impact. BMC Public Health. 2015;15:457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Hillier-Brown FC, Bambra CL, Cairns JM, Kasim A, Moore HJ, Summerbell CD. A systematic review of the effectiveness of individual, community and societal-level interventions at reducing socio-economic inequalities in obesity among adults. Int J Obes (Lond). 2014;38(12):1483–1490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Baruth M, Sharpe PA, Parra-Medina D, Wilcox S. Perceived barriers to exercise and healthy eating among women from disadvantaged neighborhoods: Results from a focus groups assessment. Women Health. 2014;54(4):336–353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Griffiths F, Lindenmeyer A, Powell J, Lowe P, Thorogood M. Why are health care interventions delivered over the Internet? A systematic review of the published literature. J Med Internet Res. 2006;8(2):e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Social Media Fact Sheet. http://www.pewinternet.org/fact-sheet/social-media/. Accessibility verified May, 15 18,
- 9. Smartphone Ownership and Internet Usage Continues to Climb in Emerging Economies. Pew Research Center; Washington DC: Jacob Poushter; 2016. [Google Scholar]
- 10. Q4-2019 MDMR: Mobile Share of Social Media Visits in the United States From 4th Quarter 2017 to 4th Quarter 2019. Columbia, MD: Merkle; 2020. [Google Scholar]
- 11. Fox S. The Social Life of Health Information, 2014. Washington, DC: Pew Research Center, The Social Life of Health Information; 2014. [Google Scholar]
- 12. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: Building an international consensus for the reporting of behavior change interventions. Ann Behav Med. 2013;46(1):81–95. [DOI] [PubMed] [Google Scholar]
- 13. Waring ME, Jake-Schoffman DE, Holovatska MM, Mejia C, Williams JC, Pagoto SL. Social media and obesity in adults: A review of recent research and future directions. Curr Diab Rep. 2018;18(6):34. [DOI] [PubMed] [Google Scholar]
- 14. Chang T, Chopra V, Zhang C, Woolford SJ. The role of social media in online weight management: Systematic review. J Med Internet Res. 2013;15(11):e262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Herring SJ, Cruice JF, Bennett GG, Davey A, Foster GD. Using technology to promote postpartum weight loss in urban, low-income mothers: A pilot randomized controlled trial. J Nutr Educ Behav. 2014;46(6):610–615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Herring SJ, Cruice JF, Bennett GG, Rose MZ, Davey A, Foster GD. Preventing excessive gestational weight gain among African American women: A randomized clinical trial. Obesity (Silver Spring). 2016;24(1):30–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Silfee VJ, Lopez-Cepero A, Lemon SC, et al. Adapting a behavioral weight loss intervention for delivery via Facebook: A pilot series among low-income postpartum women. JMIR Form Res. 2018;2(2):e18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Silfee VJ, Lopez-Cepero A, Lemon SC, Estabrook B, Nguyen O, Rosal MC. Recruiting low-income postpartum women into two weight loss interventions: In-person versus Facebook delivery. Transl Behav Med. 2019;9(1):129–134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sport Sci. 1992;17(4):338–345. [PubMed] [Google Scholar]
- 20. Contractor N. Understanding and enabling online social networks to support healthy behaviors In: Chai SK, Salerno JJ, Mabry PL, eds. Advances in Social Computing, SBP 2010, Vol. 6007. Lecture Notes in Computer Science. Berlin, Heidelberg: Springer; 2010.
- 21. Wright K. Perceptions of on-line support providers: An examination of perceived homophily, source credibility, communication and social support within on-line support groups. Commun Q. 2000;48:44–59. [Google Scholar]
- 22. Blackburn GL, Wollner S, Heymsfield SB. Lifestyle interventions for the treatment of class III obesity: A primary target for nutrition medicine in the obesity epidemic. Am J Clin Nutr. 2010;91(1):289S–292S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Kitahara CM, Flint AJ, Berrington de Gonzalez A, et al. Association between class III obesity (BMI of 40–59 kg/m2) and mortality: A pooled analysis of 20 prospective studies. PLOS Med. 2014;11(7):e1001673. [DOI] [PMC free article] [PubMed]
- 24. Keyserling TC, Samuel-Hodge CD, Pitts SJ, et al. A community-based lifestyle and weight loss intervention promoting a Mediterranean-style diet pattern evaluated in the stroke belt of North Carolina: The Heart Healthy Lenoir Project. BMC Public Health. 2016;16:732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Cavallo DN, Chou WY, McQueen A, Ramirez A, Riley WT. Cancer prevention and control interventions using social media: User-generated approaches. Cancer Epidemiol Biomarkers Prev. 2014;23(9):1953–1956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Cavallo DN, Tate DF, Ries AV, Brown JD, DeVellis RF, Ammerman AS. A social media-based physical activity intervention: A randomized controlled trial. Am J Prev Med. 2012;43(5):527–532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Centers for Disease Control and Prevention. National Health and Examination Survey (NHANES): Anthropometry Procedures Manual. Atlanta, GA: Centers for Disease Control and Prevention; 2017. [Google Scholar]
- 28. Geaney F, Fitzgerald S, Harrington JM, Kelly C, Greiner BA, Perry IJ. Nutrition knowledge, diet quality and hypertension in a working population. Prev Med Rep. 2015;2:105–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Sallis JF, Grossman RM, Pinski RB, Patterson TL, Nader PR. The development of scales to measure social support for diet and exercise behaviors. Prev Med. 1987;16(6):825–836. [DOI] [PubMed] [Google Scholar]
- 30. Wilson KE, Harden SM, Almeida FA, et al. Brief self-efficacy scales for use in weight-loss trials: Preliminary evidence of validity. Psychol Assess. 2016;28(10):1255–1264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Dean CB, Lundy ER. Overdispersion. In: Balakrishnan N, Colton T, Everitt B, Piegorsch W, Ruggeri F, Teugels JL, eds.Wiley StatsRef. 2014;1–9. [Google Scholar]
- 32. Ejiogu N, Norbeck JH, Mason MA, Cromwell BC, Zonderman AB, Evans MK. Recruitment and retention strategies for minority or poor clinical research participants: Lessons from the healthy aging in neighborhoods of diversity across the life span study. Gerontologist. 2011;51(suppl 1):S33–S45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Cavallo DN, Sisneros JA, Ronay AA, et al. Assessing the feasibility of a web-based weight loss intervention for low-income women of reproductive age: A pilot study. JMIR Res Protoc. 2016;5(1):e30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Pagoto SL, Waring ME, Schneider KL, et al. Twitter-delivered behavioral weight-loss interventions: A pilot series. JMIR Res Protoc. 2015;4(4):e123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Waring ME, Moore Simas TA, Oleski J, et al. Feasibility and acceptability of delivering a postpartum weight loss intervention via Facebook: A pilot study. J Nutr Educ Behav. 2018;50(1):70–74.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Valle CG, Tate DF. Engagement of young adult cancer survivors within a Facebook-based physical activity intervention. Transl Behav Med. 2017;7(4):667–679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Merchant G, Weibel N, Patrick K, et al. Click “like” to change your behavior: A mixed methods study of college students’ exposure to and engagement with Facebook content designed for weight loss. J Med Internet Res. 2014;16(6):e158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Napolitano MA, Whiteley JA, Mavredes MN, et al. Using social media to deliver weight loss programming to young adults: Design and rationale for the Healthy Body Healthy U (HBHU) trial. Contemp Clin Trials. 2017;60:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Edney S, Looyestyn J, Ryan J, Kernot J, Maher C. Posts, pics, or polls? Which post type generates the greatest engagement in a Facebook physical activity intervention? Transl Behav Med. 2018;8(6):953–957. [DOI] [PubMed] [Google Scholar]
- 40. Turner-McGrievy GM, Tate DF. Weight loss social support in 140 characters or less: Use of an online social network in a remotely delivered weight loss intervention. Transl Behav Med. 2013;3(3):287–294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Hales SB, Davidson C, Turner-McGrievy GM. Varying social media post types differentially impacts engagement in a behavioral weight loss intervention. Transl Behav Med. 2014;4(4):355–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Kite J, Foley BC, Grunseit AC, Freeman B. Please like me: Facebook and public health communication. PLoS One. 2016;11(9):e0162765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Rus HM, Cameron LD. Health communication in social media: Message features predicting user engagement on diabetes-related Facebook pages. Ann Behav Med. 2016;50(5):678–689. [DOI] [PubMed] [Google Scholar]
- 44. Klassen KM, Borleis ES, Brennan L, Reid M, McCaffrey TA, Lim MS. What people “like”: Analysis of social media strategies used by food industry brands, lifestyle brands, and health promotion organizations on Facebook and Instagram. J Med Internet Res. 2018;20(6):e10227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Edelmann N. 13. What is lurking? A literature review of research on lurking. In: The Psychology of Social Networking. Vol. 1. Sciendo Migration; Berlin, Boston: De Gruyter; 2015:159–174. [Google Scholar]
- 46. Samuel-Hodge CD, Johnston LF, Gizlice Z, et al. Randomized trial of a behavioral weight loss intervention for low-income women: The weight wise program. Obesity (Silver Spring). 2009;17(10):1891–1899. [DOI] [PubMed] [Google Scholar]
- 47. Davis Martin P, Rhode PC, Dutton GR, Redmann SM, Ryan DH, Brantley PJ. A primary care weight management intervention for low-income African-American women. Obesity (Silver Spring). 2006;14(8):1412–1420. [DOI] [PubMed] [Google Scholar]
- 48. Schippers M, Adam PC, Smolenski DJ, Wong HT, de Wit JB. A meta-analysis of overall effects of weight loss interventions delivered via mobile phones and effect size differences according to delivery mode, personal contact, and intervention intensity and duration. Obes Rev. 2017;18(4):450–459. [DOI] [PubMed] [Google Scholar]
- 49. Bucholz K. America’s Top New Year’s Resolutions for 2020.https://www.statista.com/chart/20309/us-new-years-resolutions-2020/. Accessibility verified February 7, 2020.
- 50. Huffaker DA. The impact of group attributes on communication activity and shared language in online communities. First Monday. 2011;16(4). doi:10.5210/fm.v16i4.3450. [Google Scholar]
- 51. Franz MJ, VanWormer JJ, Crain AL, et al. Weight-loss outcomes: A systematic review and meta-analysis of weight-loss clinical trials with a minimum 1-year follow-up. J Am Diet Assoc. 2007;107(10):1755–1767. [DOI] [PubMed] [Google Scholar]