Community-level factors may hinder or facilitate adult residents’ attempts to consume a healthful diet and be physically active.¹ For example, there are inverse associations between access to supermarkets and farmers’ markets and obesity;² and between access to recreational facilities and obesity.³ Recent review articles have found that supermarket availability was generally inversely related to obesity, and fast food availability was generally positively associated with obesity.^{4, 5} In the rural environment, distance to recreation facilities, feeling unsafe from crime, and few non-residential destinations were associated with obesity.⁶ In addition, built environmental characteristics, such as access to places to be active and neighborhood walkability are associated with physical activity (PA) and obesity.⁷ Furthermore, residents of neighborhoods that have higher Walk Score®, (a measure of neighborhood amenity density) tend to walk more compared to those with lower Walk Score®.^{8, 9}

These associations between the food and PA environment and diet and PA-related activities and outcomes have led to the hypothesis that environmental context might moderate the effect of diet and PA-related intervention outcomes. Two studies have examined the hypothesis that dietary behavior change interventions may be more effective when participants live in areas where more healthful foods are available^{10, 11} with the potential causal mechanism being that individuals who live in areas with more healthy eating opportunities are more likely to increase healthy eating behaviors. Both studies found greater adherence to dietary interventions among those with improved access to healthy food sources such as supermarkets, farmers’ markets, and green carts. ^{10, 11} Four additional studies have examined factors in the perceived and objectively-assessed built environment related to PA, generally finding improvements in PA among those participants who lived in more favorable PA environments.^12–15 However, both the diet- and PA-focused studies^10–15 were set primarily in urban areas, further supporting the need to determine whether the food and PA environment may moderate the effect of diet and PA interventions among rural residents.

Therefore, in the Heart Healthy Lenoir (HHL) Project lifestyle study, we compared changes in diet, PA, and weight loss (at 6-months) among lifestyle intervention participants who resided in healthier food and PA environments to those who resided in less healthy environments. We hypothesized that those who lived closer to supermarkets and farmers’ markets, and further from fast food restaurants and convenience stores, would have greater intervention-related increases in fruit and vegetable consumption and greater improvements in overall diet quality over the intervention period, compared to those living in less healthy food environments. We also hypothesized that those who lived closer to PA resources (e.g., parks, gyms), and in more walkable, low-crime areas would have greater intervention-related increases in total PA and walking (as assessed by steps) over the intervention period, when compared to those living in neighborhoods less conducive to PA. Our study is unique from others in that our study 1) was set in a rural environment in the Southern United States, whereas others were set in urban areas; 2) examined both perceived- and objectively-measured aspects of the food and PA environments; and 3) included both self-reported and objectively measured outcome data on intervention-related dietary and PA changes.

Table 1 shows baseline participant characteristics for those who returned for 6-month measures. The mean age of participants was 56.5 years, mean BMI was 36.3 kg/m², mean self-reported PA was 160.2 minutes per week, and mean steps/ day (as assessed by pedometer) was 4525. In addition, 80% were female, and 68% were African American. Only participants who also provided a valid physical address for geocoding were included in analyses including GIS measures (n = 191), and only participants residing in Kinston were included in analyses for crime data (n = 132). Those with valid address data self-reported more PA (P = 0.01), were more likely to report crime as a neighborhood barrier (P = 0.01), had higher Walk Score®, (P = 0.002), lower BMI (P = 0.04) and more steps as assessed by pedometer (P = 0.03). Those with crime data were younger (P = 0.02), self-reported more PA (P = 0.02), had higher diet quality as assessed by the DRA (P = 0.01), reported too many fast food restaurants and too much crime as neighborhood barriers (P = 0.04, P = 0.02; respectively), and had higher Walk Score® (P = 0.0002), compared to those without crime data.

Table 2 reports bivariate associations between food environment variables and diet-related outcomes at 6-month follow-up. There was a low, statistically significant correlation between greater supermarket density and greater improvements in the carotenoid index. There was a low, statistically significant correlation between higher mRFEI scores and weight change, indicating greater weight loss among those living in less healthy food environments. There was a positive correlation between six-month change in diet as assessed by the DRA and perceived nutrition barriers: those who perceived more nutrition-related neighborhood barriers had greater improvements in diet than did those who perceived fewer barriers.

Table 3 reports bivariate associations between PA environment variables and PA-related outcomes at 6-month follow-up, with two low, statistically significant negative correlations. The inverse associations between 1) density of private gyms and pedometer-measured steps/day indicate those living in areas with more gyms had lower increases in steps/day compared to those living in areas with fewer gyms and 2) crime and weight change indicate greater weight loss among those living in higher objectively-measured crime neighborhoods.

In Table 4, we present parameter estimates, standard errors, and P-values for multiple linear regression analyses for the six bivariate associations with a P-value < 0.10. In adjusted analyses, there was an inverse association between weight change and mRFEI, suggesting that participants who lived in a less healthy food environment lost more weight (P = 0.01). In addition, contrary to our initial hypotheses, there was a positive association between self-reported total PA change and distance to private gyms, such that those who lived further from private gyms reported greater increases in total PA from baseline to six-months (P = 0.04), and an inverse association between pedometer measured steps/day and density of private gyms, such that those living in areas with greater density of gyms had lower increases in the number of daily steps from baseline to six months (P = 0.03).

In this study, we examined the potential for moderation of intervention effects by comparing diet and PA outcomes among participants residing in healthier food and PA environments to those residing in less healthy environments. In correlational analyses, we generally found small but statistically significant improvements in diet and PA among those who lived in less favorable food and PA environments, contrary to our original hypothesis, and to previous findings in urban settings.^{10, 11, 13, 42} In adjusted analyses, which controlled for baseline values of outcomes and socio-demographic variables, findings again indicated that those who lived in less favorable environments lost more weight and increased PA to a greater degree than those in healthier, more supportive environments. Our results could suggest a hypothesis to be tested in rural settings: when resources are less plentiful, intrinsic motivation may influence an individual’s ability to make healthy changes to a greater degree than supportive environments and community resources. It could be that our intervention targeted behaviors that are easier to achieve in less favorable food environments (e.g., moderate consumption of healthy fats found in nuts, versus increasing fiber, fruits and vegetables, as the primary nutrition message), and led to improved diet and weight outcomes. Also, it may be that rural residents in disadvantaged areas are more accustomed to overcoming barriers to various goals and thus more persistent when making lifestyle changes. Finally, rural residents may be exercising in non-traditional venues (e.g., paths, churches, local schools) which were not documented in our GIS analyses.

Our findings are the opposite of Wedick et al.¹⁰ and Feathers et al.¹¹ who found that a more supportive food environment (e.g., greater access to Green Carts, farmers’ markets, and supermarkets) was associated with improved intervention-related outcomes. However, these prior studies included interventions focused on reducing dietary fat and increasing fruit, vegetable, and fiber consumption, and both were set in urban areas of the Northeastern United States. The different dietary guidance provided by the HHL Project, along with differences in geography, culture, and participant characteristics, may have contributed to our unexpected findings.

With regard to physical activity, Zenk et al.¹² found that access to indoor walking facilities was associated with improved adherence to a walking intervention among midlife African American women from Chicago. In two studies based in San Diego County, Kerr et al.¹³ found that overweight men improved walking if they lived in less walkable neighborhoods, whereas overweight women increased PA to a greater degree if they perceived better safety from traffic. Our results are again, contrary to Zenk et al. and Kerr et al.’s findings among women,^12,13 with our findings indicating that intervention participants increased PA to a greater degree if they lived further from PA resources such as gyms. Again, this could be due to differences in geography and culture of participants in our study versus other studies.

Individually-based lifestyle interventions may be particularly important for those who live in areas with less access to health-promoting community resources. In the future, in-depth interviews or focus groups could be conducted with individuals who live in less healthy environments but who overcome barriers to healthy eating and PA and achieve measurable results, to determine factors that enable participants to overcome neighborhood barriers. In addition, future work could examine whether participants who are motivated to make healthy changes are more likely to overcome neighborhood and community-level barriers to a healthy diet and PA, as well as whether those with less motivation are more sensitive to such barriers and may need special tips on how to overcome the barriers.

This study was limited in that we only were able to include participants who had baseline and 6-month follow-up data. Thus, we potentially had inadequate power to address the research questions. The sample size for some analyses was reduced because of missing data related to lack of available crime data, step data, and GIS data. An additional limitation is the possibility of obtaining statistically significant results by chance, due to examining multiple associations. The food environment variables were significantly correlated, with distance to supermarkets strongly correlated with distance to fast food restaurants (r=0.94), indicating the complex nature of the food environment variables. In addition, many studies indicate that people do not shop at the supermarket closest to home,^41,42 so it could be that those individuals who are selecting to shop at places further from home are shopping at those locations because they have healthier food options. The 1-mile buffer used in GIS analyses is likely inadequate to define participants’ neighborhoods and to encompass all trips made for food and physical activity-related activities.³¹ Finally, our generalizability was limited as our sample was enrolled from one county in eastern NC.

Prior studies in urban environments suggest individuals living in healthier food and PA environments were able to make greater dietary and physical activity improvements when enrolled in lifestyle interventions, compared to intervention participants living in less healthy environments.^10–13 Contrary to prior studies in urban locations, in our rural setting, we found that those living in less resource-rich food and PA environments generally made greater improvements in diet and PA. Our findings may be due in part to differences in the type of intervention provided, differences in how a rural environment mediates efforts to improve lifestyle behaviors, differences in motivation and self-efficacy for lifestyle change among rural residents, or other factors. Further studies should be undertaken to examine these paradoxical findings, and if confirmed, to better understand them.