摘要：Objectives. We examined whether supermarket choice, conceptualized as a proxy for underlying personal factors, would better predict access to supermarkets and fruit and vegetable consumption than mere physical proximity. Methods. The Seattle Obesity Study geocoded respondents’ home addresses and locations of their primary supermarkets. Primary supermarkets were stratified into low, medium, and high cost according to the market basket cost of 100 foods. Data on fruit and vegetable consumption were obtained during telephone surveys. Linear regressions examined associations between physical proximity to primary supermarkets, supermarket choice, and fruit and vegetable consumption. Descriptive analyses examined whether supermarket choice outweighed physical proximity among lower-income and vulnerable groups. Results. Only one third of the respondents shopped at their nearest supermarket for their primary food supply. Those who shopped at low-cost supermarkets were more likely to travel beyond their nearest supermarket. Fruit and vegetable consumption was not associated with physical distance but, with supermarket choice, after adjusting for covariates. Conclusions. Mere physical distance may not be the most salient variable to reflect access to supermarkets, particularly among those who shop by car. Studies on food environments need to focus beyond neighborhood geographic boundaries to capture actual food shopping behaviors. Socioeconomic disparities in diet quality are well established in the United States. Low income and low levels of education have been consistently linked with poor diets and poor health. 1,2 Multiple individual-level factors are known to influence diet quality, including economic barriers, inadequate nutrition knowledge and awareness, food preferences and attitudes, and cultural factors. 2–6 In the past decade, availability of healthy foods at the neighborhood level has been proposed as an environmental determinant of diet. 7–12 In the absence of data on actual food shopping destinations, it has been assumed that mere availability of supermarkets in one’s neighborhood reflects accessibility. Thus, physical proximity to neighborhood supermarkets, based on either aggregated data (e.g., the presence 11,13,14 or density of supermarkets within a given neighborhood census tract 15–17 ) or individual-level measures (e.g., the distance between one’s home and the nearest supermarket 13,16,18,19 ), has been linked with higher consumption of fruits and vegetables and with higher overall diet quality. As a result, ensuring physical access to supermarkets in low-income neighborhoods has recently become the focus of public health policies designed to improve diets and health. 20,21 However, a few recent studies have produced inconsistent results. 21–23 Some of these investigations have shown that physical distance to a supermarket is not associated with fruit and vegetable intake or overall diet quality and body weight, 22–24 even when varying distances from respondents’ homes are assessed. 25,26 Interestingly, a few studies that probed for actual food shopping locations revealed that most people did not even shop at their nearest supermarkets. 22,23,27,28 Such studies are beginning to imply that physical proximity may not be the most salient variable in determining access to supermarkets, particularly among those who shop by car. 26,29,30 Rather, it has been proposed that a host of other proximal and distal determinants of dietary intake, such as personal choices, psychosocial factors, and unobserved measures of socioeconomic status (SES), may determine supermarket choice, food shopping decisions, and, in turn, diets and health. 23,28 We examined whether physical proximity to supermarkets or underlying personal factors would more strongly predict access to supermarkets and consumption of fruits and vegetables. We were able to investigate this issue because, unlike most other studies, we had available data on actual food shopping destinations. Novel measures were used to conceptualize each of these variables. First, we measured physical proximity using street network distance between respondents’ homes and the supermarket they reported as their primary food source. Second, we used choice of primary supermarket as a proxy for unmeasured underlying personal factors such as economic and psychosocial barriers and food preferences. We hypothesized that supermarket choice would better predict respondents’ fruit and vegetable consumption than mere physical distance. The question was whether supermarket choice would outweigh physical proximity in predicting fruit and vegetable consumption even among low-income or other vulnerable groups.
