The Role of Maternal Social Networks on the Outcomes of a Home-Based Childhood Obesity Prevention Pilot Intervention

Obesity is difficult to treat, and trajectories for biological and behavioral risk factors for excess weight are established early in life (Astrup, Dyerberg, Selleck, & Stender, 2008; Birch & Davison, 2001). The stability of obesity across the lifespan, and the physical, psychosocial and financial costs related to obesity for the individual and the larger society (Finkelstein & Bilger, 2012; Finkelstein, Graham, & Malhotra, 2014; Gooze, Hughes, Finkelstein, & Whitaker, 2012; Hoelscher et al., 2015) stress the importance of correcting weight trajectories in early childhood, before children become overweight or obese (Blake-Lamb et al., 2016; Taveras, 2016). In the United States (U.S.), children from low-income families and families of color (in particular, Hispanic and African American) are at the highest risk for developing overweight, and prevention efforts in the past decades have failed to address these disparities (Ogden et al., 2016). Childhood obesity prevention needs to target broad and multilevel mechanisms that extend beyond the parent-child context to other socioecological domains, such as extended family and community social contexts, and neighborhood environments (Campbell, 2015; Skinner & Foster, 2013). Initiating and maintaining healthy habits is difficult, especially for families that face a multitude of barriers, such as poverty, housing instability, food insecurity, and a lack of access to linguistically and culturally competent services (Flynn et al., 2006). Reducing risk factors and increasing protective factors in at-risk underserved families’ social and physical environments may be critical for enabling long-term behavior change that would reduce childhood obesity and obesity inequities.

Few effective interventions to prevent obesity in early childhood exist, and existing programs typically focus on altering the behavior of a parent and/or child (Blake-Lamb et al., 2016). Parent weight and related health behaviors are important predictors of child weight and behaviors (Davison, Francis, & Birch, 2005; Wrotniak, Epstein, Paluch, & Roemmich, 2004), contributing to the intergenerational transmission of obesity risk. Family-centered obesity interventions, which target children and their parent(s), have proven more effective than interventions targeting children alone (Epstein, Paluch, Roemmich, & Beecher, 2007; Epstein, Valoski, Wing, & McCurley, 1994). However, families are themselves embedded in social networks and communities that can also influence family members’ health behaviors. For example, multigenerational family members often share similar eating and physical activity behaviors (e.g., Munsch et al., 2007; Sallis, Patterson, Buono, Atkins, & Nader, 1988); and other relationships, such as friends and neighbors, can influence weight and related behaviors (Pachucki, Jacques, & Christakis, 2011; Salvy, Feda, Epstein, & Roemmich, 2017; Salvy, Feda, Epstein, & Roemmich, 2016; Zhang, de la Haye, Ji, & An, 2018). Broader social network influences on families’ obesity and health behaviors operate through a complex set of interpersonal and socio-structural mechanisms, which include social influence, social support, and social capital (Ashida et al., 2011; Ashida, Wilkinson, & Koehly, 2012; Bauer, Neumark-Sztainer, Fulkerson, Hannan, & Story, 2011; Bauer, Laska, Fulkerson, & Neumark-Sztainer, 2011; Gorin et al., 2008). These social factors can promote or impede behavior change and intervention outcomes (Valente, 2010, 2012) and ultimately play an important role in family-based efforts to prevent obesity in early childhood.

The current study draws on social influence and social network theories to identify features of family social networks that support or hinder the outcomes of a novel early childhood obesity prevention program. This intervention, described in greater detail in Salvy, de la Haye, Galama, & Goran (2017) and de la Haye et al. (2018), involves the delivery of early childhood obesity prevention as part of the services provided by our home visiting program (HVP) partner. Annually, over 500 publicly and privately funded HVPs provide services to more than 650,000 low-income, underserved children and their families in the United States (U.S. Department of Health and Human Services, 2015). HVPs identify families pre-partum based on risk factors for poor child outcomes (e.g., low-income), and provide in-home services by trained home visitors for the first two to three years of the child’s life. HVPs are successful in addressing maternal and infant psychosocial and physical health (Fergusson, Grant, Horwood, & Ridder, 2006; Gonzalez & MacMillan, 2008; Matone, O’Reilly, Luan, Localio, & Rubin, 2012), but they have not yet targeted obesity, or comprehensively addressed maternal and infant nutrition and physical activity. In response to this gap and pressing need for programs that reach at-risk populations to prevent childhood obesity in the first years of life, our team has developed a culturally-tailored curriculum to promote healthy eating and activity habits that can be incorporated into existing HVPs. The in-home delivery of obesity prevention makes it possible to guide parents as they implement changes in their family, home, and community, and address ecologically-relevant barriers to the formation of healthy eating and activity habits. The goal of the program is to promote a healthy weight in mothers and children, and healthy eating (increased fruit and vegetables; decreased sugar-sweetened beverages) and activity (increased steps) in mothers, to reduce children’s obesity risk.

Children and adults are likely to have social connections whose risk for overweight is similar to their own (Christakis & Fowler, 2007; Koehly & Loscalzo, 2009; Zhang et al., 2018). This occurs in part because of social selection based on homophily and propinquity, whereby people are likely to establish social connections to others who share similar genetics (family), characteristics (e.g., demographics), and beliefs (McPherson, Smith-Lovin, & Cook, 2001). Similarity in weight status among social connections also occurs because obese individuals are socially marginalized, and in response are more likely to become socially connected to one another (Brewis, 2014; de la Haye, Robins, Mohr, & Wilson, 2011). In addition to these social selection processes, there is evidence that individuals’ diet, activity, and weight status are influenced by their social connections and the characteristics of their social networks; and that obesity ‘diffuses’ through networks such that an individuals’ risk for excess weight is greater when their social connections are obese (Zhang et al., 2018). Obesity appears to be influenced by our social networks because of multiple social functions that may or may not be provided by our social connections:

In sum, children and families with the highest risk for overweight and obesity are likely to be socially connected to others who are also at a higher risk for obesity, and these social environments can exert influences that are a barrier to healthy behavior change. Indeed, one study examining the influence of social networks on weight loss among Black and Hispanic adults who were overweight and enrolled in a weight loss trial, found that having obese network members who lived in the participants’ home predicted weight gain, while support from children and coworkers predicted greater weight loss (Winston et al., 2015). Obesity prevention initiatives that identify social network influences on family behavior change, and then explicitly target families’ social networks to either bolster relationships that provide positive influences, or alter connections that are barriers to change, have the potential to improve intervention adoption and effectiveness (Gest, Osgood, Feinberg, Bierman, & Moody, 2011; Valente, 2012). Close social connections that play a role in families’ day-to-day lives will be particularly important to engage, as these social ties are key sources of support and social influence on obesity-related behaviors. Eating and physical activity are daily, habitual behaviors driven more by contextual and environmental cues (including social cues), than by exposure to information or individuals’ beliefs (Gardner, 2015). Therefore, weak social ties, such as online acquaintances, may have minimal impact on promoting long-term changes to healthy eating and activity. Although online social connections may effectively diffuse information or influence simple health decisions or behaviors (e.g., Centola, 2010), a recent meta-analysis found limited evidence that online social network interventions lead to sustainable or meaningful change in key modifiable health behaviors including dietary intake and physical activity (Maher, Lewis, Ferrar, Marshall, De Bourdeaudhuij, & Vandelanotte, 2014).

Given the clustering of obesity risk in social networks and the need to engage close social ties, interventions that target communal coping processes are promising (Lyons, Mickelson, Sullivan, & Coyne, 1998), whereby health risks (e.g., obesity) are recognized as a shared risk by a social group, and this elicits interpersonal communication, support, and co-engagement in risk-reducing behaviors within the group. Family-centered interventions that have adopted a communal coping approach were found to be effective at increasing the density of health communication, encouragement, and health behavior co-engagement within family social networks, and these network changes were ultimately a mediating mechanism in healthy behavior change (Ashida et al., 2012; Koehly et al., 2011). Overall, social network intervention strategies have proven successful in other areas of public health (Valente, 2010), but to date have been underutilized in obesity prevention despite evidence that they are promising and needed.

This pilot study examines whether characteristics of mothers’ social networks are associated with the primary targets of a childhood obesity prevention initiative: namely, mother diet, physical activity, weight status, as well as infant diet and weight trajectory. Specifically, the aims were to identify how the following social network characteristics, which represent social functions that can support or interfere with behavior change, were associated with intervention targets: (a) sources of health-relevant social support, including informational support and encouragement, where we expect that greater density of support relationships bolsters healthy behavior change; (b) co-engagement in dietary and activity behavior with social ties, which may provide key opportunities for social influence and communal coping; and (c) the presence of relationships that are perceived as barriers to health behavior change. A secondary aim is to explore whether the composition of mothers’ social networks, based on the characteristics of network members (such as gender, social roles, frequency of contact), is associated with intervention outcomes.

Mothers and infants exposed to this pilot study showed trends for positive changes in healthy eating and weight (for mothers), compared to those receiving the standard HVP, suggesting that the intervention shows promise for reducing childhood obesity among at-risk children. Changes in mother and infant health outcomes were also associated with characteristics of mothers’ social networks. Mothers participating in this pilot study had social networks that were largely comprised of family members who were densely connected to one another; the majority of whom lived in the same household or neighborhood as the mother. Mothers regularly ate or prepared food with about half of these social ties and did physical activity with approximately one-quarter of ties. Overall, this suggests that the close-knit social networks measured in this study could be important sources of influence, either positive or negative, on mother and infant weight, diet, and activity, particularly via the co-engagement in these behaviors. Indeed, the results of this pilot study indicate that some characteristics of low-income mothers’ social networks may be protective (e.g., associated with better mother and infant health outcomes), while other characteristics may be barriers to health behavior change (i.e., related to poorer mother and infant health outcomes), in the context of a family-based childhood obesity prevention intervention.

Characteristics of mothers’ social networks at baseline or post-intervention that predicted one or more positive health outcomes in mothers and/or children included: (1) having a greater proportion of network members who helped or encouraged the mother to have a healthy lifestyle; and (2) having a greater proportion of social ties that were female. Social network characteristics at baseline or post-intervention associated with one or more negative health outcomes included having: (1) a greater proportion of network members who had daily or almost daily contact with the mother; and (2) a greater proportion of network members who were friends. One network characteristic was not clearly protective nor risky: having a greater proportion of network members that lived in the same neighborhood was associated with an increase in fruit intake in moms, but higher SSB intake in infants. It is difficult to interpret this last finding because we did not measure network members’ food intake. However, because health risk behaviors cluster in social networks and neighborhoods, we might expect neighborhood-based social ties to be similar in their eating habits and as such may be a barrier to behavior change unless they are also engaged in the change process.

Of the functional relationships hypothesized to be sources of social influence, only “the receipt of help and encouragement to have a healthy lifestyle, eat healthy foods, or be active” was found to significantly predict positive intervention outcomes, suggesting this may be a particularly important type of social influence needed to adopt healthy lifestyles in this population (see Figure 1). We did not find evidence that other potential sources of social influence predicted the intervention outcomes, including the proportion of network members with whom mothers ate or prepared food, did physical activity, provided health advice/information, or were perceived to be barriers to adopting a healthy lifestyle. This may be because the impact of these relationships is dependent on whether or not the person providing the information and/or influence is themselves engaging in healthy eating or activity behaviors or has a healthy weight. Unfortunately, these alter health attributes were not measured in the current study, but should be the focus of future work. Social influence theory would suggest that receiving information from, or engaging in, health behaviors with “healthy” social ties can bolster healthy behavior change, while these same relationships with “unhealthy” people may be a barrier to change.

Figure 1.

Of note, some social network characteristics appeared to have a negative influence on infant outcomes. For instance, mothers who interacted more frequently (i.e., daily) with a greater number

of their social ties, or who had greater proportion of relationships who lived in their neighborhood, had infants that consumed more SSB post-intervention and had greater increases in SSB intake over the intervention period. Frequency of contact with a greater number of ties and neighborhood proximity were also associated with greater increases in infant weight (WLZ). Having more face-to-face and neighborhood-based social ties could be a barrier to healthy changes for these mothers if the influence of these social contacts conflicts with the healthy behaviors mothers are trying to adopt (e.g., if they encourage or provide SSB for the infants). This is likely given the clustering of health risk behaviors, such as poor diet, inactivity, and obesity, within families and communities. Engaging multiple family members and close neighborhood contacts as part of interventions may be an important extension to family-focused interventions. A communal coping approach that can foster increased communication, encouragement, and co-engagement in healthy eating and activity is a promising strategy to engage multiple members of these close-knit social networks in behavior change and positive social influence (Ashida et al., 2012; Koehly et al., 2011). The benefits of targeting the family as part of lifestyle interventions are well established (Epstein, Paluch, Roemmich, & Beecher, 2007; Skelton, Buehler, Irby, & Grzywacz, 2012; St Jeor, Perumean-Chaney, Sigman-Grant, Williams, & Foreyt, 2002). The rationale underlying family-centered approaches is that modification of children’s social and physical environments is necessary to change and maintain children’s healthy habits. These findings extend the literature on the benefits of family-centered lifestyle interventions and emphasize that prevention programs should explicitly address the influence of wider social systems to help prevent unhealthy eating habits and promote children’s healthier lifestyle trajectories.

Although this pilot study has numerous strengths, including the diversity of the sample, the comprehensive measures of adiposity and associated behaviors in mothers and infants, and the use of objective measures for anthropometrics and physical activity, there are also several limitations. The small sample size limits the power to identify social network factors that predict the outcomes of interest in this six-month intervention. Additionally, because of the small scope of this study, only six social network ties were enumerated from mothers, and future work should assess a larger sample (e.g., 20) of social contacts to capture more variability in social network characteristics. The use of home visitors to gather the subjective measures (dietary intake and social network structures) is also a limitation. Overall, the results of this pilot study should be considered suggestive and verified in future larger-scale interventions.

This pilot study suggests that HVPs are a promising context for childhood obesity prevention. The results also indicate that family-based interventions seeking to promote healthy weight and related behaviors among mothers and infants may benefit from explicitly trying to increase the help and encouragement mothers receive from the people in their social networks for adopting a healthy lifestyle, eating healthy foods, or being active. Mothers in our pilot study who received more of this type of health-related support from their social networks had significantly lower BMIs post-intervention, and significantly greater intake of vegetables post-intervention. The findings also suggest that interventions may need to engage multiple social network members who have regular contact with mothers and/or who live in the same neighborhoods as mothers to help establish healthy goals and healthy lifestyles for the family, as these connections may be a barrier to change if they provide conflicting norms or messages.

There is evidence that health behavior change interventions can be designed to collectively engage social network members, elicit communal coping responses, and enhance relationships that provide health-related social support, to boost the adoption of targeted health behaviors (Valente, 2010; Valente & Pitts, 2017). Network intervention strategies to engage and elicit support from multiple social network members may also benefit family-based obesity interventions in low-income populations, to enhance protective factors in mothers’ social networks that bolster efforts to prevent childhood obesity.