This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on http://formative.jmir.org, as well as this copyright and license information must be included.
Efforts to translate evidence-based weight loss interventions, such as the Diabetes Prevention Program (DPP), to low-income postpartum women have resulted in poor intervention attendance and high attrition. Strategies that improve engagement and retention in this population are needed to maximize the reach of evidence-based weight loss interventions.
The objective of this study was to adapt a DPP-based weight loss intervention (Fresh Start) for Facebook delivery and to evaluate its feasibility among low-income postpartum women.
This study comprised 3 single-group pilot studies where feasibility outcomes iteratively informed changes from one pilot to the next. We paralleled the in-person program for Facebook delivery by translating the protocol to a content library of Facebook posts with additional posts from lifestyle coaches. Low-income postpartum women were recruited from Women, Infants, and Children (WIC) clinics in Worcester, Massachusetts. Participants were enrolled into a 16-week weight loss intervention delivered via Facebook. During the first 8 weeks, Facebook intervention posts were delivered 2 times per day, with additional posts from coaches aiming to stimulate interaction among participants or respond to participants’ questions and challenges. For the following 8 weeks, posts were delivered once per day without additional coaching. Feasibility outcomes were engagement (defined by number of likes, comments, and posts measured throughout intervention delivery), acceptability, and retention (survey at follow-up and assessment completion rate, respectively). Changes in weight were also assessed at baseline and follow-up.
Pilot 1 had a retention rate of 89% (24/27), and on average, 62% (17/27) of women actively engaged with the group each week during the 8-week coached phase. Mean weight loss was 2.6 (SD 8.64) pounds, and 79% (19/27) would recommend the program to a friend. Pilot 2 had a retention rate of 83% (20/24), and on average, 55% (13/24) of women actively engaged with the group weekly during the 8-week coached phase. Mean weight loss was 2.5 (SD 9.23) pounds, and 80% (16/24) would recommend the program to a friend. Pilot 3 had a retention rate of 88% (14/16), and on average, 67% (11/16) of women actively engaged with the group weekly during the 8-week coached phase. Mean weight loss was 7.0 (SD 11.6) pounds, and 100% (16/16) would recommend the program to a friend.
Our findings demonstrated that a Facebook-delivered intervention was acceptable and could be feasibly delivered to low-income postpartum women. Future research is needed to evaluate the efficacy of a Facebook-delivered weight loss intervention.
Obesity rates are disproportionately high among racial and ethnic minority and low-income groups [
A small body of literature has translated evidence-based interventions, such as the Look AHEAD (Action for Health in Diabetes) [
Social media (eg, Facebook, Twitter, and Instagram) holds great promise as a potential means to deliver behavioral weight loss interventions, while overcoming previously identified challenges to participation and engagement among low-income postpartum women [
In multiple studies, social media has been one of the several components of behavioral interventions focusing on diet, physical activity, weight management, smoking cessation, and sun protection [
All 3 pilot studies followed similar methodology unless otherwise noted. Each pilot study utilized a single-group pretest-posttest design. Study procedures were approved by the Institutional Review Board at the University of Massachusetts.
Study participants were low-income postpartum women in Worcester, Massachusetts recruited from the Worcester Women, Infant, and Children (WIC) program over the course of 8 months [
Intervention content was adapted from the original Fresh Start intervention, a weight loss treatment protocol based on content from the DPP [
This intervention was adapted for delivery via Facebook. First, the research team held regular meetings to review the protocol for the in-person Fresh Start Trial [
Finally, 8 research staff members with experience in weight loss intervention and social media approaches participated in a mock Facebook group where posts were delivered and pilot-tested. These individuals were asked to provide feedback on the wording of the posts, including the language, pictures, and videos, and whether the posts emphasized behavioral weight loss strategies and principles of motivational interviewing. Posts were revised and finalized for the first pilot based on feedback from the mock group.
The intervention consisted of an 8-week intervention phase followed by an 8-week maintenance phase delivered via a secret, private Facebook group, preceded by a 90-minute in-person orientation session. Each of the 3 pilots held 2 orientation sessions (one in the morning and one in the evening) that women attended based on their needs, and sessions were required for participation in the study. The orientation allowed women to meet other women in their Facebook group, provided instruction on how to join the Facebook group, and informed women about the rules of the Facebook group.
Women were also introduced to the concepts of goal setting and taught how to track their diet, physical activity, and weight loss using a commercial mobile app (MyFitnessPal) or paper records. They were also provided a scale, pedometer, workbook, measuring cups, and a 1-year gym membership to YWCA Central Massachusetts at no cost to them.
Following the orientation, participation in the Facebook intervention commenced. During the first 8 weeks, Facebook intervention posts were delivered 2 times per day (8 am and 4 pm), 7 days per week, from study-created Facebook accounts (one from the intervention coach and one from the assistant coach) via the social media management platform Buffer [
Outcome measures for each pilot study focused on feasibility outcomes including retention, engagement, and satisfaction. [
Height and weight were measured at baseline and after 16 weeks by trained research staff using a stadiometer and digital scale, respectively, with participants removing their shoes.
As in previous studies [
Survey items at the follow-up assessment asked about satisfaction with the intervention overall and satisfaction with the amount of weight lost during the program. Responses were rated on a 5-point scale ranging from “very dissatisfied” to “very satisfied.” Women also rated how helpful they found the program in helping them lose weight on a 5-point scale from “very unhelpful” to “very helpful.” We asked the participants how likely they were to participate in a similar weight loss program and how likely they were to recommend the program to a friend on a 5-point scale from “very unlikely” to “very likely.” Women also rated how often they felt supported by other participants in the group on a 5-point scale ranging from “never” to “always.” Finally, women rated the extent to which they felt the other participants were motivating and the extent to which they felt the coaches were helpful and motivating based on a 5-point scale from “strongly disagree” to “strongly agree.”
Upon completion of the postintervention assessment, participants were invited to participate in a 60-minute group discussion, during which they provided feedback on their experience in the intervention
A total of 29 women were initially enrolled in Pilot 1. However, 2 participants became ineligible during the intervention, one due to medical reasons and the other due to intake of a weight-altering medication (study exclusion criterion). Thus, the final sample for Pilot 1 was 27 women (
Baseline characteristics of women participating in the three pilot studies.
Characteristics | Pilot 1 (n=27) | Pilot 2 (n=24) | Pilot 3 (n=16) | |
Age, mean (SD) | 32.1 (5.6) | 29.4 (6) | 29.4 (4) | |
Body mass index, mean (SD) | 35.1 (5.5) | 38.2 (6) | 34.9 (7) | |
Hispanic or Latina | 11 (41) | 5 (21) | 7 (44) | |
Non-Hispanic black | 5 (19) | 7 (29) | 2 (13) | |
Non-Hispanic white | 8 (30) | 9 (38) | 6 (38) | |
Asian | 2 (7) | 1 (4) | 0 (0) | |
Other | 1 (4) | 2 (8) | 1 (6) | |
High school degree or less | 13 (48) | 7 (29) | 4 (25) | |
Some college or 2-year degree | 8 (30) | 10 (41) | 9 (56) | |
College degree or more | 6 (22) | 7 (29) | 3 (19) | |
Single | 10 (37) | 8 (33) | 7 (44) | |
Married or living with partner | 16 (59) | 14 (53) | 8 (50) | |
Divorced or separated | 1 (4) | 1 (4) | 1 (6) | |
Posted a Facebook status once per week or more | 17 (63) | 18 (75) | 11 (69) | |
Posted a video or photo to Facebook once per week or more | 16 (59) | 16 (67) | 10 (63) | |
Commented on a friend’s Facebook post once per week or more | 22 (82) | 22 (92) | 13 (81) |
Weight loss outcomes at the 16-week follow-up are presented in
A majority (75%, 18/24) of the participants reported being satisfied or very satisfied with the intervention, and 19 out of 24 women (79.2%) felt the program was somewhat or very helpful in facilitating their weight loss (
We learned several key lessons from Pilot 1 based on the group discussions and Facebook group engagement data. First, we explored intervention posts with lower engagement, as defined by <5 comments on the post or <3 women who commented on the post, to identify key themes or similarities between posts with poor engagement. From our review of engagement data, we discerned that goal setting posts, posts with lengthy videos, and posts asking multiple questions were associated with lower engagement among the participants. As a result, we edited goal posts to include sample goals, exchanged long videos for infographics, and reduced the word count of lengthier posts to simplify the posts further. Of note, when iterations were made to the approach to behavioral strategies, the core evidence-based behavioral strategies were maintained across groups. During the group discussions, we learned that women were more likely to read and comment on a post that included helpful pictures and visuals. For example, a few of the women shared that they took “screenshots” of the intervention pictures and saved them to the photo library on their mobile phones to be able to access them later. We therefore carefully reviewed intervention posts that included photos (eg, a picture of a mom walking) versus infographics (eg, a diagram of healthy snacks with calorie amounts), and where applicable, we enhanced posts to include infographics.
Second, during the first few weeks of the intervention, women reported via Facebook posts to the group and private messages to the coaches that they were having difficulties tracking their diet or activity using the recommended app (ie, MyFitnessPal). We thus modified intervention posts during the first 2 weeks to include posts that provided support in learning to use the app. Specifically, we created 4 new Facebook posts for Pilot 2 that included both videos and infographics and directed women on key features of tracking their food using the app.
Finally, during the group discussions, women reported that they wished there were more opportunities for the participants to meet in person and suggested that this be accomplished via increased utilization of the free YWCA membership. Based upon this suggestion that in-person interactions be in the form of workout groups or other exercise participation, during the coached phase of Pilot 2, participants received Facebook posts with invitations to join and participate in an exercise class at the YWCA with other women in the group and one of the coaches. Class type (eg, Zumba, yoga) and day and time varied each week to facilitate the attendance of women with different scheduling needs.
Measures of intervention engagement for the three pilot studies.
Measures | Pilot 1 (n=27) | Pilot 2 (n=24) | Pilot 3 (n=16) | ||||
Coached phase | Noncoached phase | Coached phase | Noncoached phase | Coached phase | Noncoached phase | ||
Original posts | 2 (2.5) | 0 (0.62) | 3 (3.8) | 0 (0.3) | 3 (2.6) | 0 (0.3) | |
Comments | 18 (25.2) | 2 (4.4) | 24 (31.2) | 4 (6.6) | 16 (15.5) | 3 (4.9) | |
Likes | 33 (57.7) | 5 (8.8) | 19 (27.1) | 3 (4.5) | 19 (21.8) | 2 (2.8) | |
Women who engaged in all 8 weeks | 8 (33) | 1(4) | 4 (20) | 2 (83) | 4 (29) | 1(19) | |
Women who engaged in ≥3 out of 8 weeks | 21 (88) | 11 (41) | 17 (85) | 11 (46) | 12 (87) | 6 (38) | |
Women who did not engage | 1 (4) | 10 (37) | 3 (13) | 9 (38) | 0 (0) | 6 (38) |
aEngagement indicators are averages rounded to the nearest whole number.
Weight change and participant satisfaction outcomes in the three pilot studies.
Outcomes | Pilot 1 (n=24) | Pilot 2 (n=20) | Pilot 3 (n=14) | |
16-week weight change (lbs), mean (SD) | −2.6 (8.64) | −2.5 (9.23) | −7.0 (11.6) | |
16-week weight change (%), mean (SD) | −1.4 (4.38) | −1.2 (3.97) | −3.6 (5.6) | |
16-week body mass index change, mean (SD) | −0.5 (1.5) | −0.5 (1.6) | −1.3 (2.2) | |
Lost any weight at 16 weeks, n (%) | 15 (63) | 10 (50) | 10 (71) | |
Gained any weight at 16 weeks, n (%) | 9 (38) | 9 (45) | 2 (14) | |
Satisfied with the program, n (%) | 18 (75) | 16 (80) | 9 (64) | |
Program helpful in facilitating weight loss, n (%) | 19 (79) | 12 (60) | 11 (79) | |
Would recommend program to a friend, n (%) | 19 (79) | 16 (80) | 14 (100) | |
Would continue program after study ends, n (%) | 15 (63) | 18 (90) | 13 (93) |
Pilot 2 had initially enrolled 25 participants. However, 1 woman became ineligible due to pregnancy, leading to a sample of 24 women (
An average of 13 (55.2%) women engaged in the group each week. Of the 20 women who completed the 16-week assessment, 70% (14/20) reported reading the entire intervention post either most of the time or always, and 65% (13/20) women said that they read only part of the post most of the time or always. When asked about lurking behavior, 35% (7/20) women reported occasionally reading a post without commenting on it or liking it; 25% (5/20) women reported lurking half the time, 25% (5/20) women much of the time, and 5% (120) woman always.
Women lost an average of 2.5 (SD 9.23; range −26.2 to 8.8) pounds or 1.2% of their baseline weight (SD 3.97; range −9.8 to 3.8;
A majority (16/20, 80%) of the women reported being satisfied or very satisfied with the intervention, and 60% (12/20) women felt the program was somewhat or very helpful in their weight loss effort. Furthermore, 65% (13/20) women felt supported by other women in the group at least half of the time, and 60% (12/20) women felt the other women in the group were motivating. Finally, 90% (18/20) women felt that the coaches were helpful, and 95% (19/20) women felt that the coaches were motivating.
Observations and findings from Pilot 2 provided additional insight on how to enhance the intervention. First, women endorsed opportunities to attend group exercise classes with one of the intervention coaches at the YWCA. However, only 3 of the women ever attended these classes. Challenges to attendance included lack of time, class timings not working well with their schedules, and childcare responsibilities (babysitting). Second, women expressed that they wished more women would have engaged actively in the Facebook group. Factors that got in the way of active engagement included feeling uncomfortable posting about themselves and not wanting to post if they were not experiencing successful weight loss. In terms of intervention content, women expressed that they were less likely to read posts that were too long and endorsed posts that elicited responses via questions and included visuals. Examples of helpful visuals included pictures of appropriate portion sizes, recipe substitution ideas, and number of calories of specific commonly consumed foods.
From the lessons learned in Pilot 2, we made several modifications for Pilot 3. First, we again carefully reviewed the posts based on the women’s feedback, as well as posts that elicited low engagement (<5 comments or <3 women commenting). We adjusted the language to make them more concise and engaging, and where applicable, we replaced some of the text with a picture or video. In response to feedback regarding the use of gym, we extended the orientation by 5 minutes to include more information about the YWCA and provide women with a map to the facility and the group exercise schedule. We were also able to secure babysitting at the YWCA for 2 hours, 3 days per week: 1 day in the morning, 1 day in early afternoon, and 1 day in the late afternoon or early evening. To encourage women to post more in the group, we added 2 posts in the first week of the intervention; one asked women to introduce themselves to the group and one encouraged women to post in the group regardless of their motivation throughout the program (eg, when they lost weight vs when they did not lose weight; when they had a good week vs when they had a bad week).
Pilot 3 had 17 enrolled participants. However, 1 woman became ineligible during the study due to a new pregnancy, leading to a final sample of 16 women (
On average, 67% (11/16) women engaged in the group each week. Out of the 14 women who completed the 16-week assessment, 43% (6/14) women reported reading the entire intervention post most of the time or always (≥75% of the time), and 43% (6/14) said that they read part of the post most of the time or always. When asked about lurking behavior, 36% (5/14) women reported occasionally reading a post without commenting on it or liking it; 21% (3/14) women reported lurking half the time, 36% (5/14) women much of the time, and 7% (1/14) woman always.
During Pilot 3, women lost an average of 7.0 (SD 11.6; range −31.8 to 13.8) pounds or 3.6% of their baseline weight (SD 5.62, range −11.9 to 6.1;
A majority (9/14, 64.3%) of the women reported being satisfied or very satisfied with the intervention, and 79% (11/14) women felt the program was somewhat or very helpful for their weight loss effort. Furthermore, 64% (9/14) women felt supported by other women in the group at least half of the time, and 57% (8/14) women felt the other women in the group were motivating. Finally, all 14 (100%) women felt that the coaches were helpful and 93% (13/14) women felt that the coaches were motivating.
We learned several key lessons from the women in Pilot 3. Women reported that they would have liked there to be more interaction among the women in the group, and a few suggested more opportunities for in-person meetings or get-togethers (eg, Pilates in the park, walking around the neighborhood). However, like Pilot 2, a discrepancy existed between these suggestions and attendance in in-person group opportunities, as none of the women in Pilot 3 utilized the babysitting or attended the classes at the YWCA with one of the coaches. Women also reported several factors that contributed to their engagement with the Facebook group, including the time of day of the post, how much time had passed since the post, and how many other women had commented on the post. Women also offered several suggestions for improving communication between each other within the Facebook group, such as exchanging phone numbers during the orientation session, having nonweight loss related ice breakers on the Facebook group (eg, asking about work schedules or how many kids women have), and having weekly step or weight loss competitions. Finally, women provided feedback on intervention content and reported that they found practical tips related to meal planning, proper food storage, food preparation, and recipes to be particularly helpful.
Based on the findings and feedback from Pilot 3, as well as Pilots 1 and 2, important lessons that could inform further intervention refinement were the intervention posts that included concise language; open-ended questions that elicited responses; infographics and content related to weight loss progress, calorie goals and budgeting; and practical tips for meal planning and cooking. Furthermore, while women in all 3 pilots liked receiving the intervention content via Facebook, they desired more social support from one another and suggested solutions, such as a longer orientation session, more ice breakers and discussions on Facebook, and face-to-face interactions to facilitate this support. It is important to note, however, that when women in these pilots were offered more opportunities for in-person interaction, they did not utilize them.
This series of 3 pilot studies demonstrated that an adapted behavioral weight loss intervention delivered primarily via Facebook is feasible and acceptable among low-income postpartum women. On average, more than half of the women in each of the 3 pilots actively engaged (measured by likes, comments, and posts) in the group each week during the coached phase of the intervention (weeks 1-8). Furthermore, women were highly receptive as shown by the fact that most of them found that the intervention was helpful for losing weight, the coaches were supportive, and they would be likely to recommend the program to a friend. Given this high level of receptivity, the finding that satisfaction decreased slightly from pilot 1 to pilot 3 is difficult to interpret. Finally, women in Pilot 3 lost an average of 7 pounds compared with those in Pilots 1 and 2 who lost 2.6 pounds and 2.5 pounds, respectively.
Several previous studies among postpartum women enrolled in WIC programs have observed no significant changes in body weight following a variety of intervention delivery modalities including digital virtual discs (DVDs) and support group teleconferences [
While direct comparisons with in-person interventions are not possible due to different delivery modalities, our results suggest slightly higher engagement levels. For example, in the original Fresh Start pilot intervention, 26% (7/27) of the women attended the first 3 weekly sessions, while 54% (36/67) of women across our 3 pilots participated in the Facebook group at least once in each of the first 3 weeks [
Despite a higher participation in the Facebook-delivered intervention, active engagement decreased throughout the intervention, particularly after the removal of the coach, in all 3 pilots. Instead, we observed an increase in passive engagement given that with fewer women liking, posting, and commenting, more women reported lurking. Previous research with low-income postpartum women has also consistently demonstrated poor engagement in weight loss intervention sessions [
To improve active engagement across the 3 pilots, we included additional opportunities for in-person exercise classes at the YWCA based on the women’s suggestions. However, when provided the opportunities to meet in person, women did not participate. This is an important factor for researchers to consider when evaluating available resources for interventions, and additional research is required to better understand this discrepancy. Future research may consider utilizing user experience design, which observes individuals in their daily lives, to better inform intervention design{Yardley, 2015 #1414}. This methodology may serve to close the gap between participant suggestions or desires and actual behaviors.
One approach to sustain engagement could be to enhance the coaching component of the intervention. As stated above, women across the 3 pilot studies were very receptive to the coaching aspect of the intervention, and active engagement declined considerably from the coached to noncoached phase of the study. Furthermore, while previous studies adapting the DPP have provided personalized feedback regarding weight, diet, and physical activity changes, feedback in this study was limited as it was based on women’s responses to posts rather than a thorough review of individual self-monitoring logs. Future studies may enhance the coaching component by providing more personalized feedback in addition to group-based feedback. Another approach could be to intervene with dyads or cohorts of individuals who already know each other [
This study has several limitations. First, the sample sizes of the 3 pilots were relatively small and decreased over the 3 pilots due to declining WIC enrollment during our study period (October 2016-August 2017) and study time constraints. A limitation in our measures was that we were not able to obtain objective measures of lurking behavior or directly relate changes in the types and content of posts to engagement. This suggests that women may have had greater exposure to the intervention that was not reflected in our measures of engagement, and future studies that utilize social media to deliver behavioral weight loss interventions should investigate strategies or intervention iterations specifically related to engagement (eg, types of posts, post content) to sustain participation over time. Additionally, we were unable to conduct mediation analysis to determine the influence of engagement as a mediator of weight loss given that the sample size of our pilot studies was small. This study recruited women who were regular Facebook users, which may limit the generalizability of our findings. However, we found that access to social media was not a major barrier to women participating in this study, and only 2 women were ineligible due to limited Facebook use. This is consistent with recent research suggesting that low-income and minority populations have similar technology access to other population subgroups. As of 2017, internet usage, smartphone ownership, and social media usage were similar among Hispanic (88%, 75%, and 74%), and black (85%, 72%, and 63%) adults compared with white adults (88%, 77%, and 69%) [
Social media-delivered behavioral weight loss interventions show great promise due to their high potential of reaching low-income diverse individuals, reducing intervention burden, and decreasing the cost of weight loss intervention delivery. The findings from this series of 3 pilot studies demonstrated that a Facebook-delivered intervention was acceptable and could be feasibly delivered to reach low-income postpartum women. Future research is needed to evaluate the efficacy and sustainability of delivering a weight loss intervention via Facebook.
Step-by-step depiction of the adaptation process for Facebook posts.
Collage of Facebook posts.
Diabetes Prevention Program
Women, Infants, and Children
This research was generously supported through grants from the National Institute of Minority Health and Health Disparities (1 P60 MD006912-02), the National Heart, Lung and Blood Institute Training Grant 1T32HL120823-01, and the Centers for Disease Control and Prevention (U48 DP005031-01). We acknowledge the contributions of our community partners and organizations that made this research possible: The Worcester WIC Program and our University of Massachusetts Medical School colleagues and staff (Karen Ronayne, Christine Frisard, and Linda Olsen).
None declared.