This study examines the relations between neighborhood measures with total energy intake and expenditure among a sample of young girls, using both indirect and direct assessments of the neighborhood environment.
While there is considerable evidence for the associations between neighborhood characteristics and health outcomes among adults, these findings are not necessarily generalizeable to children. For adults, the literature suggests that living closer to supermarkets may increase fruit and vegetable intake [35, 36], whereas living closer to fast-food restaurants is associated with obesity risk [37, 38]. Contrary to these findings, we observed an inverse association between the prevalence of "food and retail" destinations and total energy intake among young girls. Compared to adults, children may be less influenced by environmental food availability because children usually have limited ability to purchase food without supervision. Fisher and Birch  found that restricting food access among young children promoted overeating when the restrictions were lifted. On the other hand, perhaps diversification of the neighborhood food environment deters energy consumption among children due to the unrestricted accessibility to assorted food options.
Our findings show crude inverse associations between physical activity and neighborhood deprivation, access to "mixed residential and commercial" destinations, and "physical disorder." These corroborate previous studies which found similar negative associations between children's physical activity and neighborhood violent crime  and physical and social disorder .
The significant interaction effects of race/ethnicity indicate that certain neighborhood characteristics are more strongly associated with physical activity among children of different racial/ethnic groups. For example, Hispanic/Latina girls' physical activity behaviors were positively associated with neighborhood "recreation" and "walkability." "Physical disorder" was also associated with lower physical activity behaviors among African-American girls. This may result from wider variation in these scales among different racial/ethnic groups, enhancing the ability to detect associations.
A considerable strength of this study was the use of street audit data, a method generally recognized as an accurate representation of current neighborhood conditions. By using EFA, we were able to use the underlying correlation pattern of numerous street items to develop our neighborhood scales. However, the EFA model is provisional and we encourage attempts at replication. The literature for the relationship between neighborhoods and children's dietary intake is still growing; our study suggests that neighborhood food and retail availability may influence children's dietary behaviors differently from that of adults. In contrast, the associations we observed between neighborhood measures and children's physical activity behaviors corroborate previous studies using different methods of neighborhood assessment.
One limitation of this study is its cross-sectional nature, which makes it difficult to infer causation. While most neighborhood studies also employ cross-sectional designs, longitudinal studies may better capture the fluidity of the neighborhood environment and how changes over time affect the health of its residents. On the other hand, many neighborhood aspects are fairly stable within the time frame of a few years, although changes in residence would also influence neighborhood characteristics.
Second, our sample size is relatively small compared with other studies. Davison and Lawson  observed that significant associations between the neighborhood environment and children's physical activity behaviors were found mostly in studies with 1,000 participants or more, suggesting that these associations are small and only detectable within large sample sizes. While the sample size limited our ability to validate the neighborhood scales, other studies have found similar associations between the neighborhood environment and children's physical activity using indirect neighborhood assessment methods. Investigators who are conducting direct observations could use these new scales to guide their data collection.
Estimates of energy expenditure were derived from girls' individual activities; however, values based on MET equivalents for adults' activities were used to estimate total energy expenditure among girls since MET equivalents for children's activities were not available. Because all girls were classified the same way, nondifferential misclassification would not affect the overall rankings of girl's energy expenditures relative to one another, though our effect estimates for energy expenditure might be attenuated. Estimates of energy intake and expenditure were derived from self- and maternal-reported data. Though under-reporting of dietary intake is possible, study dieticians used the multiple pass dietary assessment and collected several 24-hour recalls to improve energy intake estimates. It is unlikely that systematic measurement error would occur in self-reported food intake or physical activity patterns by neighborhood characteristics; therefore, crude associations may be attenuated but not otherwise biased.
There may be unmeasured confounding by parental behaviors and practices. Under the assumption that parents can select into more desirable communities (i.e. communities with low crime rates, good public schools, parks and recreational spaces), parental practices may be a common cause of the neighborhood environment and girls' energy intake and expenditure. Though we adjusted for parental income in our analyses, residual confounding may still exist. Future studies examining associations between the neighborhood environment and children's health behaviors should collect information on parental perceptions of the neighborhood to better control for any potential confounding.
The participants from the CYGNET Study may not be representative of the general population. All participants were members of KPNC at birth and at time of recruitment, although their health coverage in the interim has not been examined. The study sample also tended to be from higher socioeconomic status households. On the other hand, study participants are diverse racially and ethnically, with over 25 percent being Hispanic/Latina, reflecting the diversity of the San Francisco Bay Area. While non-representativeness suggests caution in extrapolating inferences to the general population, the aim of this study was to determine whether variations in neighborhood measures are associated with variations in energy intake or expenditure. As indicated by the substantial variation in the number of street segments that were observed per girl, we successfully achieved variation in the neighborhood measures.
Because the study population was comprised of young girls, we cannot generalize our findings to boys. Previous studies that have examined the neighborhood environment in relation to energy expenditure have been inconsistent, with some studies showing stronger associations in girls [42, 43], and others showing larger effects among boys [44, 45]. Further research is warranted to determine if neighborhood environments affect energy intake and expenditure differentially in girls and boys. Like other neighborhood studies, we also did not measure how participants used their neighborhood environments, nor did we measure the school environment. Therefore, we cannot determine if the observed associations between neighborhood measures and dietary and physical activity behaviors are direct or resultant from unmeasured characteristics. However, such information is now being collected in follow-up interviews with CYGNET Study participants, enabling future investigation of these aspects with the neighborhood measures presented here.