This study helps elucidate some of the behavior changes in children that resulted from Shape Up Somerville, one of the first multi-level, multi-setting community-based childhood obesity prevention interventions [15
]. Baseline data indicate that there was a great need for effective programming to help children meet behavioral recommendations related to obesity prevention. Compared to controls, children in the Somerville intervention decreased sugar sweetened beverages by more than a 12-ounce can per week compared to controls and increased their participation in organized sports and activities by 0.2 per year, a slight but significant effect. They decreased their overall screen time by nearly 15 minutes per day. These behavioral changes likely helped contribute to the effect on BMI z-score that was observed in the Shape Up Somerville project [15
], as they would have helped reduce the energy gap between calories consumed relative to calories burned over and above those needed for normal growth and development by approximately 56 calories per day (a decrease in intake of approximately 144 kcals per week from sugar-sweetened beverages; and an increased expenditure of approximately 250 kcal per week from sports participation and from replacing screen time with moderate activity [33
We did not observe an intervention effect on fruit and vegetable consumption. Other studies published to date that used a multi-level, social ecological approach collectively indicate that this approach can be effective at increasing fruit and vegetable consumption whether the primary target is the home, school, or community environments [16
]. However, parent/caregiver report was likely to capture consumption that took place mainly in the home, and may have missed changes in other environments. The amount of fruits and vegetables served at school lunch did increase significantly [28
], for example.
The SUS intervention resulted in a decrease of 2 ounces of sugar-sweetened beverage per day. In addition to targeting individual behavior change through the in-school and after school curricula, environmental changes limited the availability of sugar-sweetened beverages through enactment of the wellness policy, which required beverages provided for snack in the classroom, sold as a la carte snacks, or sold for fundraisers to meet nutritional guidelines that limited sugar content. In addition, the home environment was targeted through parent nutrition forums and newsletters that raised awareness about the potential health detriments caused by sugar-sweetened beverages. In the community environment, restaurants were required to offer low-fat dairy as an alternative to sugar-sweetened beverages to become a Shape Up Somerville approved restaurant [29
]. The APPLE intervention [36
], that likewise used a multi-level approach included curriculum lessons highlighting the negative effects of sugar-sweetened beverages, increased availability of water at schools, and provision of a community-wide healthy eating guide. APPLE resulted in a decrease in intake of carbonated beverages at 2 years [16
]. This suggests that this behavioral target is a particularly feasible for modification in this type of intervention.
There was no significant intervention effect on active transport to and from school, despite substantial efforts to encourage it. Parents indicated that safety concerns were a major barrier in the formative phase of the study. Efforts to address this included the institution of walking school busses, traffic calming tactics, repainting of cross walks, and creation of maps highlighting safe routes to school. Walking was promoted through walking contests and the observance of International Walk to School Day. It appears that substantial environmental change along with awareness campaigns were insufficient to address the major barrier of safety concerns, perhaps because of the young age of the children and the highly urban environment.
We observed a significant intervention effect on the number of organized sports and physical activities per year, such as lessons and teams that children participated in. Those most frequently reported were swimming, dance, and soccer. The increase is notable since it requires a community-wide approach: more programming must be available, availability must be communicated, and barriers to participation must be removed. The intervention included a built environment training that emphasized the importance of a safe and accessible environment with good programming. It also included trainings by the SUS taskforce that worked throughout the intervention period to develop and implement a wellness policy that included increased activity opportunities. Finally, this result suggests that the Physical Activity Resource Guide was useful, and that parent/caregiver outreach efforts were successful.
Shape Up Somerville children decreased their overall screen time by nearly 15 minutes per day compared to children in the control communities. For this behavioral target, effectiveness may have been achieved through the consistent messaging that children received from their parents, teachers, after school staff, doctors, the mayor (who was a key “community champion”), and other influential adults within the community. An effect on screen time was similarly observed in the Switch [19
] and Travis County CATCH [35
] studies. This suggests that messaging implemented at multiple levels may create synergistic effects that positively influence children’s screen-related behaviors.
We did not find an intervention effect on whether a child had a television in the bedroom or not, or whether the family ate dinner in a room with the television on. Intervention components focusing more intensively and specifically on the home/family environment may be necessary to achieve change in these outcomes.
This study has a number of potential limitations. Implementation using CBPR required leveraging an established relationship with the target community. For that reason, Somerville was chosen for the intervention rather than being randomized. However, control communities were chosen to match Somerville closely based on demographic characteristics. The presence of controls helps rule out the possibility that the observed changes were the result of secular trends alone. It is also possible that the results are not generalizable to other communities. However, Somerville is a diverse urban community that had access to a typical level of resources. The intervention was designed to be flexible and to operate through settings that would be common to any community.
The behavioral outcomes were measured by parent/caregiver report, limiting the ability to capture changes that may have occurred outside the home environment. Furthermore, it raises the issue of recall bias. In particular, awareness and buy-in to the intervention may have caused parents/caregivers of children in the intervention group to perceive and report greater changes in their children’s behaviors than actually existed. While this cannot be ruled out, significant effects were not realized for all outcomes; and in fact outcomes that were heavily emphasized in the intervention, such as walking to and from school, were not significant. Recall bias due to straightforward memory inaccuracies would be expected to be similar in both control and intervention groups.
Although the intervention was designed to target the entire community, only a subset of children in Somerville and the control communities were measured and followed, and fewer still had parents/caregivers who provided complete data for the 2 years of the intervention. In all communities, enrollment in the overall study was limited by the requirement for parental informed consent in diverse communities with many languages spoken and a lack of familiarity with research. The analytic sample of parents who completed the Family Survey Form pre and post-intervention was likely further hindered by this issue, as suggested by the fact that a higher percentage of Family Survey Form responders spoke English as the primary language, despite all parents/caretakers receiving the survey in their primary language. It is possible that those who responded to study recruitment were more interested in diet and physical activity behaviors, and were already practicing healthier behaviors at baseline. For this reason, true changes may have been more difficult to detect. Finally, a limitation is that this study describes data collected in 2003 and 2005. However, the targeted behaviors remain on the national agenda since children continue to fail to meet recommendations. The social ecological and systems-based approach taken in this study remains highly relevant for childhood obesity prevention, and as more communities are taking this approach it is important to gain an understanding of the complex systems in place and the impact on behavior.
Despite these limitations, this study provides evidence for change in several targeted behaviors that was sustained over a two-year period, which included the transition from a researcher/community partnership in the first year to the community alone during the second year. The City of Somerville has continued and expanded many of the initiatives, and community-generated data suggest that children’s weight status outcomes have continued to improve in the years since the original intervention [37
]. These results therefore suggest success in building community capacity and support the efficacy of a social ecological and systems-based approach for promoting sustainable change for childhood obesity prevention.