Background
Evidence suggests there are racial/ethnic disparities in physical activity (physical activity). Additionally, individuals’ physical activity has been associated with several neighborhood characteristics such as poverty, prevalence of recreational facilities, land use, street connectivity, residential density, and safety. However, only one study has examined neighborhood characteristics as correlates of racial/ethnic disparities in physical activity. Furthermore, when examining physical activity, most examine only exercise-related physical activity, not physical activity that includes all bodily movement. Finally, few have examined a wide variety of neighborhood characteristics in relation to physical activity.

Objectives
We addressed three questions:

1) Do racial/ethnic disparities in physical activity exist and, if so,
2) Do they vary geographically and, if so,
3) Which of twelve neighborhood characteristics correlate with that geographic variation, controlling for individuals’ characteristics?

That is, are racial/ethnic disparities in physical activity stronger in some neighborhoods and weaker in others, and what might explain that?  

Method
This paper defines physical activity two ways: 1) Meeting guidelines versus all others and 2) Being at least moderately active (being moderately active or meeting guidelines) versus all others. Physical activity was measured by asking respondents about the number of minutes and days in a typical week that they did vigorous or moderate exercise or performed activities for at least ten minutes without stopping. Persons were then classified as: a) Meeting national guidelines for physical activity put forth in Healthy People 2010, b) Not meeting guidelines but being moderately active, or c) Being sedentary because of no or minimal activity. We used individual-level data from the two most recent Los Angeles County Health Surveys (1999 to 2000, n=8,354 with a 55% response rate and 2002 to 2003, n=8,167 with a 59% response rate). These were population-based, telephone surveys of random samples of adults. The analytic sample for being at least moderately physical activity included only persons (n=12,901) providing accurate cross street data so they could be assigned census tracts. However, we were only able to use 2002-2003 data (n=6,405) for the other dependent variable (meeting guidelines) because the two surveys did not ask comparable physical activity items. Neighborhoods were operationalized as census tracts (n=1,630). The eleven neighborhood variables of interest were: two measures of land use, two measures of street connectivity, residential density, perceived safety, number of arrests, percentage of land space that is park space, racial/ethnic composition, racial/ethnic diversity and mean household income. We conducted cross-sectional, random effects (multi-level) analysis of individuals nested within neighborhoods. Random effects analysis is useful when individuals are nested within larger groups, such as neighborhoods, because it adjusts standard errors to account for the lack of independence among residents of the same neighborhoods. We controlled for gender, education, age, hours worked, income, marital status, physical disability, smoking status, self-reported general health status (fair/poor versus good/excellent) and having a chronic health condition (asthma, diabetes, heart disease or depression).

Results
There were racial/ethnic disparities in both types of physical activity. Asian Pacific Islanders (APIs) were half as likely as Whites to meet the physical activity guidelines and almost half as likely (OR=0.6) to be at least moderately physical activity. There also was a trend for Latinos (p=0.06) to be more likely to be at least moderately physical activity than Whites (OR=1.2). The disparities for both physical activity measures varied across census neighborhoods (i.e., were stronger in some neighborhoods than others). When we simultaneously added tracts’ residential density and individuals’ perceptions of neighborhood safety to the model predicting guideline-meeting physical activity, the disparities disappeared. Similarly, when we simultaneously added percentage of park space to the model predicting moderate or better physical activity, the disparities disappeared. However, these results show that the average resident of neighborhoods with higher residential densities were more likely to meet the physical activity guidelines and those perceiving their neighborhoods to be unsafe were less likely. The average resident of neighborhoods with higher percentages of park space were more likely to have at least moderate physical activity. A second set of multivariate analyses showed that, relative to Whites, APIs lived in areas with lower proportions of park space and higher residential densities and were more likely to think their neighborhoods unsafe.

Conclusions
Neighborhood characteristics may partially explain racial/ethnic disparities in physical activity. Living in areas that are more densely packed with housing may increase one’s ability to meet the physical activity guidelines because residents may walk more often for utilitarian or social purposes in these more urban areas. Having more park space nearby may increase the likelihood of pleasure walking or exercising, as parks can be a destination point and an opportunity for uninterrupted physical activity.

Support
Center for Population Health and Health Disparities, 1P50ES012383-01, National Institute of Environmental Health Sciences.