Participants were recruited from a large academic medical center in Southern California that draws patients including those residing in areas with some of the highest proportion of Latino individuals in the United States. Using data available in patient medical records, potential participants were identified as (1) female, (2) aged at least 18 years, (3) diagnosed with stage 1 to 3 breast cancer, (4) having completed active treatment, (5) having a BMI greater than or equal to 25 kg/m² and less than or equal to 43 kg/m² at study entry, and (6) self-identifying as either White and Hispanic or Latino and/or Spanish speaking. We excluded individuals for whom moderate physical activity was contraindicated and those with metastatic disease, recurrent cancer, visual or hearing impairment, or any major psychiatric and/or life-threatening illness that precluded their ability to consent. Potential participants were randomly selected to receive a letter from their health care provider describing the study and were contacted by telephone by study personnel. Through our institution’s National Cancer Institute–designated cancer center, we identified 3261 (n=617, 18.9% Latina) women who had been diagnosed with breast cancer and completed active treatment. A total of 275 women were contacted and 44 (16%) ultimately consented to participate in the study. The CONSORT (Consolidated Standards of Reporting Trials) diagram of our recruitment process is shown in Figure 1.

Study procedures were approved by the Institutional Review Board at the University of California, Irvine (2014-1543). All participants provided informed consent. Data were deidentified. Participants were compensated for participating in the study.

After providing consent, participants were stratified by ethnicity and randomized into an intervention or active comparison group: either Mi Vida, Mi Salud or Mi Salud, respectively. This was done via a form of adaptive randomization, minimization (1:1 ratio), because this was a small study and simple randomization could result in covariate imbalances [37]. The characteristics used for group assignment were age, time since diagnosis, stage of disease, whether patients had chemotherapy and/or radiation, and current use of aromatase inhibitors or tamoxifen [38]. All participants were given a smartphone preloaded with the corresponding mobile app. In Mi Salud, women recorded diet, exercise, fatigue, mood, sleep, and pain. Mi Vida, Mi Salud contained these same features as well as tools for self-discovery. The self-discovery feature presented the participant’s own data back to them through a series of filters. For example, participants were able to filter information pertaining to their diet, exercise, weight, or sleep by their mood or pain levels. An example of this is shown in Multimedia Appendix 2: in this image, word clouds show the association between what a participant has eaten and their symptoms, with red indicating fatigue and green indicating energy. The greater the size of the font, the more that food had been consumed. Therefore, the app allowed participants to assess the association between their behaviors and symptoms. Participants used the apps for 12 weeks from 2017 to 2018.

The apps were culturally and individually tailored, as they were available in both Spanish and English, and participants could record their decisions about food and exercise choices. For example, participants could record their mood and then rate its intensity on a scale.

At 3 time points (baseline, 6 weeks, and 12 weeks), we collected biometric and survey data from participants; however, we only focused on data collected at baseline and 12 weeks. We measured participants’ weight and height to calculate their BMI.

Surveys included items assessing health-related social support, health-related social control (eg, persuasion and pressure), and undermining [39]. Three items were used to assess health-related social support [32,40]. A sample item included, “Over the past month, how often did the important people in your life show appreciation for your efforts to stay on track with your diet or exercise regimen?” (0=not at all; 5=every day). The composite measure of health-related social support demonstrated good reliability in this sample at baseline (Cronbach α=0.90) and 12 weeks (Cronbach α=0.87) [32,40].

Seven items, 3 to assess persuasion and 4 to assess pressure, were derived from prior research to assess health-related social control [32,40]. Participants were asked to indicate the frequency with which their social network members sought to exercise health-related social control in the past month. A sample item for persuasion included, “Over the past month, how often did the important people in your life try to do something to get you to improve your food choices or exercise regimen?” A sample item for pressure included, “Over the past month, how often did the important people in your life restrict you from making poor food choices or from being physically inactive?” Responses were made on a 6-point scale (0=not at all; 5=every day), and composite variables were created to represent each strategy of health-related social control. Items pertaining to pressure and persuasion at baseline (Cronbach a=0.87 and Cronbach a=0.87, respectively) and at 12 weeks (Cronbach a=0.68 and Cronbach a=0.85, respectively) also had good reliability.

Health-related undermining was assessed with 7 items from the sabotage subscale of the Family and Friends Support for Heart Healthy Eating Habits scale [39]. Sample items included “refused to eat the healthy food you prepared” or “offered you high-sugar, high-fat foods.” Responses were made on a 6-point scale (0=never; 5=always). The composite measure demonstrated good reliability in this sample (Cronbach α for undermining at baseline and 12 weeks were α=0.79 and α=0.79, respectively).

At time 1, 22 non-Hispanic White and 22 Latina participants provided complete survey data. At time 2, of the 22 participants in each group, 19 non-Hispanic White (86.4%) and 18 (81.8%) Latina women provided complete data at the 12-week follow-up. Non-Hispanic White women used the mobile app more often than did Latina women, although this was marginal (non-Hispanic White: 60.5% of days vs Latina: 37.7% of days; P=.09); however, mobile app use data were available from only 18 (81.8%) Latina and 13 (68.4%) non-Hispanic White participants, indicating that more Latina participants actually downloaded and used the app compared with non-Hispanic White participants.

We used Fisher exact test and 2-tailed t tests to compare differences in demographics, changes in BMI, and changes in social network processes among Latina and non-Hispanic White participants, from baseline to 12-week follow-up. We then used linear regression to assess the association between group (1=Mi Salud, Mi Vida and 0=Mi Salud only) and BMI change score. We then used 4 separate linear regression models to assess the association between change in BMI as the outcome and change in social network process as the predictor. We first controlled for group, ethnicity (1=Latina, 0=non-Hispanic White), BMI at time 1, and social network process score at time 1. We then added terms showing the interaction between ethnicity and the change in the social network process. Models with significant interaction terms were then further probed by calculating the predictive margins of change in BMI according to change in social network processes among Latina and non-Hispanic White participants separately. Regression models used full information maximum likelihood estimation to address missing data at the 12-week follow-up. All analyses were conducted in Stata (version 16; StataCorp LLC) [41]. We reported significance at P≤.05 and a trending association at P≤.10.

This study assessed the association between social network involvement and change in BMI among Latina and non-Hispanic White breast cancer survivors in the context of a mobile health weight loss intervention. While not the primary focus of this manuscript, we note that there was not an association between change in BMI and participation in the intervention (under review), which mirrors the findings of other similar studies [42,43]. This may be an indication that inclusion of self-discovery in the intervention group was not substantially different from the functions provided within the active control group. Nevertheless, our entire sample experienced weight loss over the course of the 12 weeks.

Our findings build upon prior work assessing the association between different social network processes and health-related outcomes. There was no main effect of change in health-related social support on weight loss in the entire sample; however, increased health-related social support was associated with decreased BMI, particularly among non-Hispanic White women, highlighting a pattern of findings consistent with the literature [44]. The lack of association between social support and weight loss among Latinas was surprising given that prior research has shown a significant increase in health-related social support among Latina women with type 2 diabetes engaged in a weight loss intervention [23,45]. The degree to which social support is experienced may vary as a function of perceived illness burden. For example, some evidence suggests that Latina breast cancer survivors in particular express concerns about burdening their families [46], and perhaps this reluctance to seek or receive social support makes this type of network provision less effective for Latina women [46].

Regarding health-related social control, we found that persuasion was not associated with change in BMI. Although multiple studies have found this relationship, this pattern aligns with other work showing that persuasion is not always effective in promoting behavior change [23,31]. In contrast, pressure was associated with weight gain, which is consistent with prior work showing the negative effects of pressure in decreasing adherence to health behaviors [32,47,48]. Our study is among the first to demonstrate an association between increased pressure and weight gain.

Finally, we found that undermining was not associated with change in weight. Prior work has shown that the effects of undermining may vary according to health behaviors: in one study, undermining of physical activity was not associated with weight change, while undermining of healthy eating was associated with weight gain [48]. Future studies are needed to examine how undermining influences weight change among breast cancer survivors.

This pilot study has limitations. As this study was designed to obtain preliminary evidence regarding the feasibility of the Mi Vida intervention, it did not include more extensive follow-up assessments, thus limiting a more detailed analysis of the potentially reciprocal process of interpersonal involvement with weight loss efforts. A second limitation was the lack of consideration of individual goal setting. Prior work has shown that monitoring progress toward a set goal is an important aspect of attaining that goal [49]. Although all women were asked to lose between 1 and 2 lb (0.45 to 0.91 kg) per week, allowing for individual autonomy in determining a personal weight loss goal was likely to improve weight loss outcomes and also provide a clear goal that could be communicated to meaningful others. A third limitation of our study was that the measures of social network involvement did not identify the specific sources of the health-related social support, control, and undermining. Future research is needed to use finer-grained measures to evaluate how changes in the behavior of specific social network members over time relate to health outcomes.

Nevertheless, this pilot study is among the first to report changes in social network interactions within the context of a weight loss intervention among breast cancer survivors. The inclusion of a large sample of Latina women is notable, given their increased risk for symptom burden and comorbidities, coupled with the lack of focus on examining social network involvement to increase healthy behaviors in this population. These findings can be used to inform future larger randomized controlled trials aimed at improving health behaviors among Latina breast cancer survivors.