Reading Journal

Environment and Planning A 2013, volume 45, pages 142 – 158

doi:10.1068/a45130

Too much food and too little sidewalk? Problematizing

the obesogenic environment thesis

Julie Guthman

Division of Social Sciences, University of California, Santa Cruz, 1156 High Street, Santa Cruz, CA 95064, USA; e-mail: [email protected] Received 16 March 2012; in revised form 3 May 2012

Abstract. The obesogenic environment thesis is that increased prevalence of obesity is because people are surrounded by cheap, fast, nutritionally inferior food and a built

environment that discourages physical activity. This thesis has animated various planning,

advocacy, and educational interventions to address these obesogenic qualities. However,

studies designed to test the thesis have generated inconclusive or marginal results, and

the more robust fi ndings may be based on spurious correlations. Part of the problem

is methodological: researchers embed many assumptions in their models and derive

causality from unexamined correlation. As such, they neglect the possibility that features

of the built environment may be as much an eff ect of sociospatial patterning as a cause.

In addition, in embedding taken-for-granted assumptions about the causes of obesity—

namely, the energy-balance model—these studies foreclose alternative explanations,

including  the  possible role of environmental toxins. This approach to studying the

obesogenic environment is a textbook example of problem closure, in which a specifi c

defi nition of a problem and socially acceptable solutions are used to frame the study of

the problem’s causes and consequences.

Keywords: energy balance, food deserts, obesogenic environments, obesity

“Data Show Manhattan is Svelte and the Bronx is Chubby, Chubby” read a headline in a July 2009 edition of The New York Times. The story reported on a study that had just been released that had compared obesity rates in the Bronx and Manhattan boroughs of New York City (Chan, 2009). In keeping with well-established inverse correlations between socioeconomic status and weight status, Manhattan’s rates of overweight and obesity were far lower than those in the Bronx, and “the prosperous swath of Manhattan from the Upper East Side down to Gramercy Park had the lowest obesity rate (less than 15 percent) in the city.” As reported by The New York Times, the head researcher, Andrew Rundle, noted that at the neighborhood level, socioeconomic and demographic factors were the strongest predictors of obesity rates. He then equivocated, stating that even when adjusting for poverty and race, at least three factors are associated with reduced obesity: proximity to supermarkets and groceries where fresh produce is sold; proximity to parks; and access to public transportation, which reduces reliance on cars. The authors thus concluded that increasing the number of produce markets and making neighborhoods more walkable might reduce obesity rates.

This study is one of dozens, and possibly hundreds, of studies completed in the last decade or so that are based on the theory that people are obese because they are surrounded by cheap, fast, nutritionally inferior food and a built environment that discourages physical activity. This theory was fi rst formalized in the academic literature as the ‘obesogenic environment’ thesis (Hill and Peters, 1998; Swinburn et al, 1999). As stated by Hill and Peters (page 1371),

Problematizing the obesogenic environment thesis 143

“ our current environment is characterized by an essentially unlimited supply of convenient, relatively inexpensive, highly palatable, energy-dense foods, coupled with a lifestyle requiring only low levels of physical activity for subsistence. Such an environment promotes high energy intake and low energy expenditure.” Along with generating research, the thesis has animated various planning, advocacy, and

educational interventions to address the obesogenic qualities of the built environment. These have included creating outlets for fresh fruits and vegetables in ‘inner-city’ food deserts, redesigning (or remarketing) public spaces to encourage walking and bicycle riding, and city-sponsored educational campaigns to achieve obesity reduction (Herrick, 2007). The last has been a favorite of cities pinned as being some of the United States’ most obesogenic, such as Houston, Texas, which topped a list created by Men’s Fitness magazine several years in a row (Herrick, 2008; Sui, 2003).

The focus on the built environment in explaining and attempting to prevent obesity is in certain respects salutary. Deemphasizing individual behaviors would seem to diminish the moral scrutiny and invocations of personal responsibility that typically accompany discussions of obesity’s causes. Moreover, it brings some focus to food industry and regional planning practices which potentially assigns culpability to powerful and malignant actors. So I do not want to dismiss this line of argument altogether. However, studies in this vein have as a whole generated inconclusive or marginal results, and the more robust fi ndings may be based on spurious correlations. Furthermore, there is little evidence that the interventions to which the thesis leads work (Evans et al, 2012). The problem is in part methodological: studies are limited by the availability and commensurability of useful data to test the thesis, and tend to rely on questionable proxies. But it is also conceptual. Owing in part to the inability of quantitative research to answer questions of causality, these studies embed several untested assumptions about the character of the problem. In this paper, I focus on two such assumptions: (1) that the built form determines peoples’ eating and exercise behaviors; and (2) that peoples’ eating and exercise behaviors determine obesity. Assumption (1) is particularly salient in discussions of the relationship between neighborhood environments, race/class, and obesity, in which higher rates of obesity among people of low socioeconomic status are explained by the type of environments they inhabit. Assumption (2) is more generally pervasive and taken for granted. It rests on the energy-balance model: obesity is assumed to result from an excess of calories taken in relative to those expended. Although many researchers acknowledge that obesity has a genetic component in addition, in this research context they generally dismiss its importance with the understanding that genetics cannot explain the abrupt rise in obesity since 1980 (Crossley, 2004).

Taken together, these critiques suggest a problem of problem closure. Problem closure refers to the situation when a specifi c defi nition of a problem is used to frame subsequent study of the problem’s causes and consequences in ways that preclude alternative conceptualizations of the problem (Hajer, 1995, page 22). It may entail embedding assumptions about a scientifi c object’s character into the research of that object (Jasanoff, 2004; Reardon, 2005). For example, the obesogenic environment thesis thoroughly embeds the energy-balance model in its assumption that it is high energy intake and low energy expenditure that the environment is responsible for. Problem closure can also entail defi ning the cause of the problem in relation to socially acceptable solutions (Forsyth, 2003). Studies of the obsogenic environment do this as well. By focusing on the built environment as a cause of obesity, they suggest that certain supply-side solutions, such as farmers’ markets and bike paths, will reduce obesity. Such solutions are often more politically palatable and doable than those that might be raised by alternative conceptualizations of the problem, that might, for example, address signifi cant income inequality, class-related stresses, or the pervasiveness of toxins, all of which may also

144 J Guthman

have a role in obesity. For that matter, defi ning obesity to begin with as a bad thing that must be prevented or stopped is an example of problem closure, as well, although that discussion is beyond the scope of this paper (but see, for example, Colls and Evans, 2009; Evans, 2006; Gard and Wright, 2005; Guthman, 2011) .

Accordingly, in this paper I discuss these fi rst two aspects of problem closure in detail, primarily by analyzing characteristic research on obesogenic environments, and then touching on research that suggests alternative conceptualizations, including the emerging research that suggests noncaloric etiologies. It will not be an exhaustive review, however. There are several published review articles, some of which discuss the character and fi ndings of studies quite systematically, and there is no need to replicate them here (eg, Black and Macinko, 2008; Booth et al, 2005; Cummins and Macintyre, 2006; Giskes et al, 2011; Townshend and Lake, 2009). So, rather than selecting a representative sample of studies, I have selected studies drawn from a variety of North American, British, and Antipodean contexts based on their ability to illustrate key methodological issues that exemplify problem closure. I pay specifi c attention to those that invoke the built environment as a cause of obesity but do not actually measure relationships between the environment and weight status. It should be noted, however, that my empirical points are based largely on North American examples which, because of the race/class segregation typical of North American cities and regions, are particularly confounding for the thesis (Cummins and Macintyre, 2006; Pearce et al, 2007; Pearce et al, 2009). Because I tack back and forth, I have endeavored to describe features of the North American context that may differ from other regions. I begin with some defi nitional work and discuss how researchers might go about studying the obesogenic environment.

Defi ning and operationalizing the (US) obesogenic environment As implied above, the obesogenic environment thesis as it relates to the built environment contains two core claims: one about the ubiquity of affordable, fast, junk food, relative to fresh, ‘healthy’ food (energy intake); and the other about the dearth of opportunities for physical activity (energy expenditure). Literature on the US food environment has tended to treat two different kinds of spaces as obesogenic. As with the UK, one is the so-called food desert. The term food desert is generally used to describe urban neighborhoods with a paucity of supermarkets and other venues at which to purchase healthful fruits, vegetables, meats, and grain products, often coupled with an abundance of liquor and convenience stores where only snack food and highly processed, ready-to-eat meals can be purchased (Cummins and Macintyre, 2002). In the US, food deserts are primarily inhabited by African Americans and some recent immigrants (Short et al, 2007). The other problematic food environment is the endless strips of ‘big-box stores’ (large warehouse-style discount stores) and fast-food and chain restaurants in which supersizing/value-meal practices fi gure prominently. It is defi ned by excess, not scarcity. Such strips tend to be located in newer suburbs, and these new suburbs tend to be inhabited by working-class and middle-class whites, although not exclusively so.

These two types of environments tend to be the culprits in the physical activity part of the thesis as well. When researchers hypothesize the obesogeneity of blighted urban cores, they suppose that low-income, dense, urban neighborhoods inhibit physical activity, based on the supposition that people might fear walking in their neighborhoods (Booth et al, 2005; Lee, 2006), although others have suggested that the fear of walking might be offset by the necessity for walking among car-less residents (Poortinga, 2006). Suburban sprawl fi gures into obesogeneity in apposite ways—the poor connectivity of street networks that increases trip distances, suburban layouts that make walking and cycling impractical and/or unsafe, the reduced viability of public transportation, and the insuffi ciency of park development are all imagined to reduce exercise activities (Plantinga and Bernell, 2005).

Problematizing the obesogenic environment thesis 145

With these defi nitions in mind, it is worth considering how researchers might go about trying to demonstrate a relationship between obesity and features of the built environment. Most turn to two established scientifi c methods for establishing relationships between health outcomes and place: geographic information systems (GIS) and spatial analysis. These involve the use of spatial statistics and mapping to demonstrate correlative relationships between places with higher obesity prevalence and environmental features that might contribute to obesity. Both approaches, however, must rely on available data to make these correlations—data which themselves lead to certain kinds of explanations and not others.

First, researchers would need to ascertain variations in the prevalence of obesity across space in order to establish that some neighborhoods, places, and/or regions have higher obesity prevalence than others. They would tend to use body mass index (BMI) as a measure of obesity, since height and weight are the size measurements collected for large numbers of people. In the United States, these are collected through the National Health and Nutrition Examination Survey (NHANES), the Behavioral Risk Factor Surveillance System (BFRSS), and various state-level surveys. The BFRSS samples many more people, but the NHANES is considered more accurate because it includes in-person interviews and medical examinations—and collects more detailed (and longitudinal) data about socioeconomic status and behaviors that can be used as variables in an analysis. Thus, the choice of which survey to use would infl uence the depth and breadth of the fi ndings. Alternatively, researchers could use self-reporting, as have many studies (Booth et al, 2005). To show that BMI values vary across space, they would need to sort individual BMI values by geocodes—codes that identify the individual with a particular state, county, postal code, and/or census tract. To measure neighborhood environmental infl uences, researchers would want these codes to be available at fi ne-grained scales but they might fi nd that, due to sampling issues in health surveys (especially detailed ones such as NHANES), it is diffi cult to obtain statistically reliable measures of BMI at fi ner grained scales than the state, metropolitan area, or county. They might then map this variation to identify clusters of high obesity prevalence.

Thus far, however, the analysis would only have identifi ed geographic variation in obesity—or, perhaps, clusters of obesity for further study. Many researchers do not begin with the map, though: instead, they identify a place or region they wish to study for its obesogeneity; or compare two places in close proximity, as in the New York study. Either way, the next step would be to identify statistical associations between higher obesity rates and environmental features. The researchers would need to hypothesize what features might actually contribute to obesity and turn them into variables. If they were testing the obesogenic environment thesis, they would want to ascertain differences in availability of good-quality and bad-quality food, and opportunities and obstacles to physical activity. To show that these statistics are relevant as geographic phenomena, they would want their models to incorporate spatial dimensions such as proximity to or density of various features. On the food side, researchers might thus be interested in the proximity, density, and mix of grocery stores, fast-food restaurants, big-box stores, and so forth. On the activity side, they might look at the availability, proximity, and safety of parks, the number and proximity of gyms, as well as ‘walkability’ more generally.

But it would not be enough to conceptualize features of the built environment that contribute to obesity. The researchers would need to fi nd data to approximate those features. They would likely turn to business censuses to obtain data on, for example, the number of restaurants, gyms, and box stores found in a geocode; to remote sensing technologies to fi nd operational measures of urban density; and to health surveys to obtain data on factors such as drive-to-work times. Researchers might also do their own surveying, perhaps walking and driving to estimate travel times to different sorts of business. Or they could forego these

146 J Guthman

‘objective data’, as they are often referred to in this research, and ask research subjects about their perceptions about these qualities of the built environment. These are typically the sources of data most published studies have used to measure these typical variables (Papas et al, 2007).

Finding data would not be the only challenge researchers face. They would also have to make their data geographically commensurate, which might entail aggregating all of their data, including their BMI data, to the largest geographic scale used. Too coarse a scale, though, would provide very little information. To get that more fi ne-grained analysis, researchers might make compromises in other dimensions of the research. For example, the North American Industry Classifi cation Systems used in the business census does not differentiate fi ne dining from family-oriented chain restaurants in the ‘full-service restaurant’ category. Whether that restaurant is a Denny’s (a national budget chain in the US) or the upscale Charlie Trotter’s in Chicago would seem to matter greatly. Many studies use variables that are not sharply defi ned in this way.

In actuality, few projects include all of these steps or a comprehensive array of variables. In the understandable interest of feasibility, most bite off a certain chunk, use variables that approximate environmental features, forego variables for which data are lacking or incommensurate, and otherwise reduce noise by focusing on relatively few elements. This perhaps is one of the reasons why in the aggregate the results of these sorts of studies have been marginal (Cummins et al, 2007). Even when they are robust, however, they do not necessarily prove what they are intended to prove.

Studying the studies Until recently, studies relating different ‘foodscapes’ to obesity have found inconclusive or inconsistent evidence, even in highly localized settings (Black and Macinko, 2008; Cummins and Macintyre, 2006; Pearce et al, 2009). A recent review of mostly US studies claimed more support for the thesis. Four out of fi ve studies that measured differences in weight status by access to supermarkets found that people with greater access to supermarkets had lower BMI and/or were less likely to be overweight or obese compared with those with less access, and fi ve out of eight studies examining access to takeaway food outlets in relation to weight status found that greater access was associated with greater BMI and/or prevalence of overweight/ obesity. However, in both sets the differences were small to moderate in magnitude. Moreover, associations of access to grocery and convenience stores were found to be mixed, as were access to restaurants and cafes (Giskes et al, 2011). Studies that have looked at the physical activity side have fi ndings that are somewhat more robust. Those that relate sprawl and weight status have been consistently conclusive, although, as I will argue, this may be a spurious correlation. In studies that measure neighborhood walkability in relation to weight status a relationship seems to be found—but not always for the reasons researchers hypothesize. Some of the marginality of fi ndings can be explained methodologically. I focus on three concerns.

Untested assumptions about the relationship between environmental characteristics and obesity In Cummins and McIntyre’s (2006) review of studies, they stated that the relationship between low-income neighborhoods and lack of good grocery stores or prevalence of fast food is well established, at least in the United States if not elsewhere. Yet, as they noted, fewer studies have actually measured that in relationship to obesity. To be sure, one of the most surprising features of this area of research is that many studies do not actually measure obesity in place or space, although still assume the tenets of the obesogenic environment thesis.

One study, for example, claimed to look at the relationship between convenient access to fast food and the prevalence of obesity among black and low-income populations, focusing

Problematizing the obesogenic environment thesis 147

on Orleans Parish, Louisiana (Block et al, 2004). The research team cited several studies that had found strong relationships between income, race, and fast-food consumption, but noted that few studies had looked at these relationships at the ecologic (spatial) level. Wanting to employ such data in their study, they found that fast-food restaurants were associated with predominately black and low-income neighborhoods, with race having stronger associations than income. Yet their study merely asserted that convenient access to fast food explains the prevalence of obesity among black and low-income populations. Even a study that invoked obesogenic environments in its title looked only at the relationship between the density of fast- food restaurants relative to socioeconomic indicators at the ecological level (Hemphill et al, 2008). Studies in the New Zealand context that show robust relationships between indices of social deprivation and the proximity of fast-food outlets make similar extrapolations about the effects on obesity (Pearce et al, 2007).

In a similar vein are studies designed to test the effects of grocery stores on diets. Several studies have found positive associations between diets composed of more fruit and vegetables and the availability of grocery stores in nearby areas (eg, Cheadle et al, 1991; Morland et al, 2002). Morland et al, for example, found that African Americans consume one-third more fruits and vegetables for every additional supermarket found in a census tract. They also found that wealthy and predominately white areas had more supermarkets and fewer neighborhood grocery stores. The study did not actually test obesity prevalence, however, although it has been cited repeatedly in the literature on the obesogenic environment. It is worth noting, in addition, that not all researchers agree on the obesogenic quality of specifi c kinds of stores. For instance, most assume that big-box stores are a feature of the obesogenic environment, based on the supposition that they encourage people to buy more food than they need for the week. Yet, in one study researchers assumed that box stores allow people to purchase fresh fruits and vegetables more cheaply (Courtemanche and Carden, 2010).

Some studies that do incorporate obesity measures have asserted a relationship even when the actual fi ndings are marginal. On the physical activity side, Rosenberger et al (2005) examined the relationship between recreation supply and obesity in West Virginia (which has some of the highest rates of obesity in the US) and found no statistically signifi cant relation- ship between the quantity of recreation opportunities and rates of obesity across various counties. They did, however, fi nd a relationship between the quantity of recreation oppor- tunities and rates of physical activity—which they then stated could explain differential rates of obesity. In short, many studies peg one set of correlations (between features of the built environment and neighborhood deprivation) to another (between low socioeconomic status and obesity), without actually examining the correlations that are at the center of their claims.

Coarse variables to test environmental features One great limitation of these studies is the absence of appropriate data with which to test possible hypotheses about the environment (Curtis and Riva, 2010). While health surveys are available to measure obesity rates in space (although often with insuffi cient numbers in rural areas), features of the built environment must almost always be approximated. As such, it appears that data availability has driven much of the research, and sometimes the data available are quite coarse (Booth et al, 2005).

For example, a study of Erie County, New York found a statistically signifi cant positive relationship between women’s BMI and diverse land use, especially when restaurants dominated nonresidential land use. But it was not possible to identify the type of restaurant that predicted higher BMIs (Raja et al, 2010). Several studies have relied on people’s perceptions of the built environment, rather than attempted to quantify the environment ‘objectively’. A complex multivariate study in Leeds, UK, sought to measure the relationship between variables related to socioeconomic status, ‘social capital’, dietary and physical activity, and

148 J Guthman

obesity among children (Procter et al, 2008). The study looked at these patterns for the city as a whole as well as by ward, fi nding, as is common, that obesity rates were patterned by socioeconomic status throughout the city. As with other studies, the strongest statistical relationship was between high socioeconomic status and fruit and vegetable consumption. Yet, how that related to features of the built environment was not well established. The data used to approximate environmental characteristics were ascertained by subjects’ responses to questions about whether supermarket access was hard or easy, access to leisure facilities was bad or good, and public transport was bad or good. While expected relationships held at the global level, the fi ndings were somewhat anomalous at more fi ne-grained scales—a problem if one is trying to make claims about neighborhood infl uences. The researchers suggested that the problem lies with ‘local infl uences’ not captured in the data, but did not entertain the possibilities either that perceptions were not adequate to the task, or that the relationship simply does not hold. A study carried out in Glasgow, testing the relationship between access to physical activity and weight status, was also based on perceptions. Controlling for socio- demographic variables, researchers found a minor relationship: in this case adolescents who reported convenient physical activity facilities were 2% less likely to be overweight/obese and 5% less likely to be obese (Nelson and Woods, 2009).

Sometimes the data are so coarse that the fi ndings veer toward spurious correlation. One oft-cited study found a relationship between numbers of fast-food restaurants per square mile and rates of obesity (Maddock, 2004). However, this was based on statewide obesity rates—a measure too global to test the signifi cance of the built environment in daily life. Likewise, another study found a correlation between the rise of obesity between 1981 and 1990 and the rise in the number and density of neighborhood stores selling sweets, pizza stores, small grocery stores, and fast-food restaurants in four towns in agricultural areas of California (Wang et al, 2008). Many things have changed in rural California during that period (including suburban sprawl, increased undocumented immigration), and it seems a major supposition to attribute growth in obesity to one set of changes. In general, data availability limits the claims that can be legitimately derived from these sorts of studies.

Untested assumptions about cause from correlation Despite the issues discussed thus far, some studies have yielded robust results. Nevertheless, the robustness is in association—not in establishing cause and effect. For instance, two studies, one in Alberta, Canada and one in California, employed a retail food environment index (RFEI). The RFEI was calculated as the ratio of the availability of fast-food restaurants and convenience stores to grocery stores and produce vendors around respondents’ homes within a certain radius. A higher index thus meant that more of the venues deemed obesogenic were closer to peoples’ homes. Both studies also controlled for neighborhood and individual socioeconomic status. Both studies found a positive correlation between higher scores on the index and obesity prevalence, with about four percentage points more in places with a ‘5’ RFEI compared with a ‘3’ (Babey et al, 2008; Spence et al, 2009). Other studies have found signifi cant associations between the likelihood of being obese in neighborhoods with a high density of fast-food restaurants in comparison with those with a low density (Li et al, 2009). These studies thus assume that proximity or density of fast-food restaurants is decisive in people’s eating habits and that, effectively, living in certain neighborhoods made people bigger than they otherwise might have been. They do not consider the factors that determine where restaurants locate (perhaps where they will have likely customers) or the possibility that people who are already obese and have low socioeconomic status move to areas with high fast food density because, for example, the real estate is cheaper.

A similar analysis can be made regarding the relationship between obesity and features that approximate possibilities for physical activity. Several studies have shown modest to

Problematizing the obesogenic environment thesis 149

signifi cant relationships between sprawl (measured by lower density/area developed) and/or vehicle miles traveled in relation to obesity (Ewing et al, 2003; Frank et al, 2003; Lopez, 2004; Lopez-Zetina et al, 2006; Vandegrift and Yoked, 2004). Lopez, for example, used individual-level data from the BFRSS with an index of urban sprawl for various metropolitan areas and found modest to signifi cant correlations between sprawl and overweight/obesity after controlling for variables such as gender, age, race/ethnicity, income, education, and even diet. Likewise, several studies have found that obesity rates are lower in mixed-use areas and neighborhoods of high walkability (based on higher residential density, mixed land use, street connectivity, aesthetics, and safety) (Li et al, 2008; Lopez, 2004; Lovasi et al, 2008; Saelens et al, 2003). The assumption underpinning these correlations is that people are made bigger (or smaller) in relation to the opportunities for physical activity in their environments. Even those who toy with reversing the direction of causation still assume that place determines behavior or weight status. Lopez (2004), for example, asked whether some people might choose to live in areas of sprawl so as to avoid walking, but then dismissed that possibility as being unreasonable. Plantinga and Bernell (2005) suggested that consumers make a trade-off between weight gain and low-price housing. Specifi cally, they claimed that people who choose low-density suburbs do so to maximize price utility whereas those who “have a stronger aversion to weight” seek out “healthier locations” (page 490).

Importantly, not all studies have concluded that obesogenic features are determinative of weight status. A study that failed to fi nd a link between BMI and restaurant proximity found a robust correlation between BMI and those who ate at those restaurants (Jeffery et al, 2006). The authors pointed out that even though density of fast-food restaurants may vary, access to fast food in the US is so ubiquitous that it had to be a matter of personal choice. Likewise, one study found that personal factors (eg, time, injury) and neighborhood barriers (eg, traffi c, unattended dogs), had stronger associations than did sprawl with physical activity levels (Joshu et al, 2008). Along similar lines, Rosenberger et al (2005) speculated that areas high in natural amenity and recreation would attract what they call “healthier migrating populations”. These and others have suggested that it may be the personal characteristics and behaviors of people who happen to cluster in space that are the basis of the correlations between obesity and space or place.

To sum up so far, the research that associates environmental factors and weight status is incomplete and inconclusive. Some attribute this to weak research design (Booth et al, 2005; Brug et al, 2006; Schafer Elinder and Jansson, 2009). Brug et al reviewed 297 observational studies and found that few results of investigations of environmental associations have been replicated, and that most studies failed to apply multilevel analyses. Such approaches are also hampered by available data: some proxies of the built environment are hard to quantify, much less collect on a postal code basis, to include in a spatial model (Cummins et al, 2005; Smyth, 2008). The meager empirical evidence for place effects relative to individual characteristics also owes to limited, conventional representations of place and space (Cummins et al, 2007). No studies have looked, for example, at unemployment rates, housing prices, proximity to cultural centers and institutions of higher learning—data that are available at the ecological level and could reasonably be used to draw statistically signifi cant associations with weight status.

Others point to the limitations of quantitative research in understanding causality. Herrick (2008) and Bagwell (2011) have separately argued for the need to attend to local context, which cannot be deduced from spatial variables. For instance, Bagwell found that the density of fast-food restaurants in the Tower Hamlets area of London was a response to demand for Halal food and an alcohol-free space by the local Bangladeshi community. In general, quantitative studies of the obesogenic environment pay scant attention to how these

150 J Guthman

environments actually affect human behavior, because they are methodologically incapable of doing so (Lake and Townshend, 2006).

Still, even those who consider more localized causes of obesity, often detected through more qualitative research, still assume that the built environment plays a determinative role. One issue which this line of reasoning neglects is that many places where thin people cluster also have features of the built environment which are considered obesogenic. For example, some of the counties with the lowest obesity rates in California are also characterized by low housing density, long commute times, and large malls (California Health Interview Survey, 2006). The difference is that these counties are comprised of relatively wealthy people. In other words, it may be socioeconomic status, not the built environment, that predicts weight status clustering in space, especially in environments where there is signifi cant race and class segregation. The neglect of this possibility suggests that the obesogenic environment thesis has not been informed by the ‘context or composition’ debate that has taken place in the literature on health geography (see Curtis and Riva, 2010; Smyth, 2008; Tunstall et al, 2004). The heart of this debate is whether it is the attributes of places or the characteristics of people who happen to inhabit them that best predict health outcomes. Given the close association between weight status and socioeconomic status, this potential rebuke of the thesis deserves consideration.

The raced and classed environment Implicit in most of these studies is the argument that different environments can explain the close correlations between socioeconomic status and weight status. Specifi cally, people of low socioeconomic status become bigger because they live in neighborhoods with fewer recreational amenities and an abundance of cheap, fast food. Yet, there are problems with this line of reasoning as well. For one, it neglects the possibility that weight status is a cause of class status rather than a consequence of it (Julier, 2008; Kirkland, 2010). This claim is based on the many studies that have demonstrated that weight bias affects people across the life course, including student–teacher relations, college admissions, marital prospects, and job advancement (Puhl and Brownell, 2001). Kirkland and Julier’s point is not that thinness guarantees high status: it is that fatness pretty much guarantees low status. If this is the case, people who are obese are more likely to locate in places where real estate is cheaper—not as a rationally chosen trade-off, but as a necessity. More generally, this corollary of the thesis ignores what makes environments ‘obesogenic’ in the fi rst place—and that, too, has to do with class and race.

Contemporary geographers emphasize that spatial patterns in housing, commercial development, and public land access are a refl ection of social relations of race and class, rather than a producer of them (Schein, 2006). These spatialized patterns of race and class have been accentuated in an era when many economic development opportunities stem from the buying power and taxability of local residents (Massey and Denton, 1998). Consider the origins of the two kinds of urban environments associated with obesogeneity discussed herein: the food deserts and suburban sprawl. In the United States, the existence of food deserts is rooted in racist insurance and lending practices (redlining), which have historically made it diffi cult to develop and sustain businesses in certain areas (Eisenhauer, 2001). Importantly, the food-desert phenomenon is also attributed to white fl ight and the net loss of supermarkets to suburbs with larger sites, fewer zoning impediments, and customers with higher purchasing power (Alwitt and Donley, 1997; Cotterill and Franklin, 1995). Conversely, much contemporary suburban sprawl has also been driven by developers and mortgage bankers encouraging a struggling, debt-ridden, middle class to move far from the urban core to take advantage of the cheap housing in areas with lower land values. These expansive working/middle class suburbs also owe much to new waves of regional economic

Problematizing the obesogenic environment thesis 151

development in the form of box stores, malls, and outlet centers—driven by the localities starved for tax revenue that encouraged such retail development to generate sales-tax revenue (Schrag, 1998). In other words, to the extent that some places have many features that are supposedly obesogenic, this often refl ects the fi nancial resources of those who inhabit such places and the waves of investment/disinvestment that have produced such environments. So the prevalence of obesity in sprawling, working-class suburbs may have less to do with low housing density and how that directly affects physical activity and more to do with who lives in those suburbs in the fi rst place.

The point can also be argued from the obverse. Much of the research pinpoints mixed-use areas as places where thin people live. Yet these gentrifi ed urban cores that contain upscale and thus, presumably, healthier eating venues as well as ample public space amenable to walking are themselves products of particular economic development strategies to attract capital (Smith, 1996). Indeed, to the extent that towns and cities with artistic, independent, and healthful restaurants, beautiful outdoor amenities, vibrant public space, and unique character are ‘leptogenic’ (causing thinness), it is because places with wealth both attract businesses to meet the food tastes of residents and generate the taxes needed to improve and maintain those enjoyable public spaces. Yet the more wealth they attract, the more they become inaccessible to many, as home prices follow. So, if thinness is a requisite for higher class status, this means that those who can afford to live in these leptogenic environments are almost necessarily thin. Perhaps this is why elite, often older suburbs escape study and thus the critique of sprawl.

What I am suggesting is that what may be ‘predicting’ the prevalence of obesity in certain places is, in fact, already existing (but unexamined) bodily differences associated with race and class, with features of the built environment, as much an effect of that spatial patterning as a cause. Rather than suggesting that composition trumps context, however, I am concurring with Cummins et al (2007) who have called the ‘composition and context’ debate a false dualism, in recognition that “there is a mutually reinforcing and reciprocal relationship between people and place (page 1835). Indeed, as Tunstall et al (2004) argue, it is precisely the inseparability of composition and context that leads to confusion over causal direction and possibly explains how area effects tend to be minor in spatial models, as has been the case in the obesogenic environment literature reviewed here.

Still, there is another possible reason that the obesogenic environment thesis has yielded results that are modest at best: this has to do with its fundamental reliance on the energy- balance model of obesogenesis. As I discuss in the next section, this too is an assumption that bears further scrutiny.

Tipping the energy balance “ It is . . . paradoxical that obesity is so persistent and diffi cult to treat, because, in Western countries at least, the basic causes of obesity are readily apparent to everyone (eating too much and exercising too little).”

Swinburn and Egger (2004, page 736)

These original authors of the obesogenic environment thesis are not alone in holding the energy-balance model axiomatic, allowing for some variation in obesity related to genetic predisposition. Even when researchers show skepticism about the operationalization of the obesogenic environment thesis (eg, when they acknowledge that fast food is so ubiquitous that thinness needs explanation), they do not budge on this most primary presumption. Yet, it appears that the energy-balance model has met some challenges, as well.

Empirically, the assumption that since 1980 people have increased the number of calories they take in relative to those they expend has simply not been demonstrated. In an exhaustive review of the literature on caloric intake and expenditure, Gard and Wright (2005) found no defi nitive proof that food intake in industrialized countries has risen and activity levels have

152 J Guthman

declined, particularly in the years since 1980. As they reported, the literature on food intake is quite contradictory, with some studies suggesting even a reduction in energy intake over the past several decades. More to the point, the degree and direction of such changes are just not known, especially in the absence of good longitudinal data. Epidemiological data are based on recollection and food diaries, which tend to underreport food intake. Supply measures, such as food-availability indices, rely on estimates of farm production, adjusted for exports and imports, nonfood uses, and food wastage (the amount of food never eaten but discarded) (USDA, 2002). These are gross estimates at best. Nor has the putative decline in physical activity been convincingly demonstrated. Studies of physical activity are also based on notoriously unreliable self-reporting (Baillie-Hamilton, 2002; Gard and Wright, 2005).

Additionally, some of the available cross-sectional data (from studies that compare behaviors across populations) debunk assumptions that certain groups eat more calories than others—the corollary of the thesis. For example, data published by the USDA on “What we eat in America”, based on national health surveys, show remarkable similarity among racial groups with respect to daily caloric intake in 2007–08: 2198 calories for whites, 2095 for African Americans, and 2109 for Mexican Americans (USDA, 2010). The same study did not show that caloric intake signifi cantly differs by income either. Those earning less than $25 000 per year reported consuming 2104 calories per day, while those earning more than $75 000 per year reported consuming 2238 per day. These surprising data may be written off to self-reporting. Yet, to dismiss such comparative data would seem to assume that certain groups are more prone to underreporting food intake than others.

Furthermore, a signifi cant array of emerging research on the biological etiology of obesity at least complicates the energy-balance model and potentially topples it as the primary factor determining weight status. For instance, disruption in sleep patterns is associated with increased body fat and altered metabolism (Bray and Young, 2007). Within nutrition science there is increased recognition that the timing and pattern of food intake affects weight gain and loss, with earlier diurnal eating leading to faster metabolism (Arble et al, 2009). Researchers are also positing that chronic stress plays a role in obesity through constant release of the hormone cortisol (Björntorp, 2001). Work in epigenetics—a relatively new fi eld that considers the role of psychosocial and nutritional stress on gene expression—has shown transgenerational effects on phenotype, including weight status (Dolinoy and Jirtle, 2008; Faulk and Dolinoy, 2011; Kuzawa and Sweet, 2009). Some in this fi eld have thus theorized that long-term exposures to stress and past nutritional deprivations could help explain the relationship between low socioeconomic status and obesity prevalence that many assume is a consequence of the current day dietary habits of the relatively poor (Thayer and Kuzawa, 2011).

Perhaps most signifi cant is research that points to the role of environmental toxins in contributing to the rise in obesity. Crucially, this research points to biological pathways to obesity that are almost entirely independent of calories (as opposed to pathways that affect mechanisms of internal regulation and thus interact with caloric metabolism) (Baillie- Hamilton, 2002; Grun and Blumberg, 2009). The most paradigm-shifting research is about the role of endocrine-disrupting chemicals. Both animal and laboratory experiments have found that maternal exposure to a range of chemicals can alter genetic pathways for fetuses in ways that generate adult obesity. For instance, both low and high doses of synthetic estrogens given to mice during gestation and immediately following birth have resulted in signifi cantly higher body weight at adulthood in progeny than that of genetically identical control groups fed the same diet. Demonstrating an epigenetic effect, this research has found that the genes that direct fat distribution are permanently altered by the exposure, such that the tendency toward high amounts of fat tissue would be passed onto offspring were the mice to reproduce

Problematizing the obesogenic environment thesis 153

(Newbold et al, 2008). Other studies have found that certain chemicals stimulate the growth of already existing fat cells as well as the development of fat cells from stem cells—those with undefi ned destination (Grun et al, 2006; Masuno et al, 2002). The epidemiological data, while thus far inconclusive, have provided some empirical support for how these mechanisms might affect humans [see Hatch et al (2010) for an overview]. For example, scientists in North Carolina found that children exposed to higher levels of PCBs (polychlorinated biphenols) and DDE (a breakdown product of DDT) before birth had higher BMIs than those exposed to lower levels (Gladen et al, 2000). It is worth noting that people with lower socioeconomic status tend to be exposed to more toxins in their workplaces and homes, particularly some of the agricultural chemicals that have been identifi ed as probable obesogens.

These studies embed their own assumptions, of course, and can be critiqued both ontologically and ethically for their reliance on genetically identical laboratory animals. So it would be folly at this point to suggest that this science presents a higher truth. Nevertheless, such fi ndings do cast doubt on the strength of the energy-balance model that is the very basis of the obesogenic environment thesis. As put by two leading scientists in this research, “the existence of chemical obesogens in and of themselves suggests that the prevailing paradigm, which holds that diet and decreased physical activity alone are the causative triggers for the burgeoning epidemic of obesity, should be reassessed” (Grun and Blumberg, 2006, page S54). In other words, there are other possible explanations for the rise of obesity and for variations in obesity related to socioeconomic status. Yet, this research is entirely outside of the frame of the obesogenic environment thesis. At some level, of course, it is to be expected. Researchers on the obesogenic environment are working in entirely different epistemic communities and, to their credit, are attempting to understand social–political dimensions of obesity. It nevertheless appears that researchers are so wedded to the energy-balance model that when results are inconclusive they tend to look for methodological shortcomings rather than question the model itself. The problem is that in neglecting these other possible causes, which also have social–political dimensions, they reinscribe a thesis that also has social– political consequences. That is the risk of premature problem closure.

Conclusion: beyond the built environment and the supply side One of the reasons I began this article with the Bronx–Manhattan study is that it differs in emphasis from many others. First, it nods to the importance of studying thinness as well as fatness: that to make any claims about neighborhood effects on obesity we have to show neighborhood effects on thinness. Second, it acknowledges that class and race have a role in explaining neighborhood differences in obesity rates, independent of the built environment. Where it falls short is in assuming that the problem features of the built environment exist independently of who lives there. I have suggested, instead, that the relationship between the built environment, spatial variation in obesity, and spatial variation in race and class are all of a piece. Gentrifi ed urban cores such as Gramercy Park are thin and wealthy, and it is unclear which begets which. Conversely, features associated with obesogeneity are precisely what make the Bronx affordable and thus available to those whose class status may exist by virtue of being obese. Yet, the quantitative spatial research that attempts to demonstrate the relationship between the built environment and obesity cannot account for this inseparability, which leads to marginal, or sometimes less than credible, results. Much more damning to the thesis is evidence that points to other possible causes of obesity having little to do with eating and physical activity, much less the built environment. Current studies of the built environment cannot possibly account for these other possible environmental explanations because they embed the energy-balance model. Researchers in this area might consider these reasons for less-than-robust results.

154 J Guthman

What is equally signifi cant about the Bronx–Manhattan study is its suggestion that obesity in the Bronx might be ameliorated by increasing the number of produce markets and making neighborhoods more walkable. This is the other aspect of problem closure: not only foreclosing other explanations, but defi ning problems in relation to socially acceptable solutions. Arguably, many of these studies work with the supply-side data, not only because they are more readily available, but also because they speak to imaginable interventions. In general, the obesogenic environment thesis, with its focus on access and proximity to grocery stores, restaurants, parks, gyms, and public transportation, leads to the conclusion that, if these conditions are changed, behaviors will follow and body sizes will transform. Other kinds of data might suggest deeper cultural and economic causes of bodily difference that are far less tractable. Arguably, that is the reason that public health professionals and food- system advocates have embraced the thesis. Supply-side interventions are relatively palatable politically and provide clarity about what to do—whether asking corner liquor stores to sell fruits and vegetables or creating more walkable public space in new suburbs. Yet, if obesogenic environments are as inseparable from race and class as I contend they are, picking out particular features of the built environment and making them more leptogenic is unlikely to be effi cacious. In effect, these are attempts to make obesogenic environments more like the kind in which thin people live, without questioning whether people were made thin by living there. Such an approach also neglects that the very conditions and amenities that make certain places sites of ‘the good life’ also make them unobtainable to most. Furthermore, championing such environments can only contribute to their economic valorization (and the reciprocal devaluation of obesogenic environments) with the real potential to exacerbate some of the inequalities they are designed to redress.

These two paradigmatic aspects of problem closure exist at our peril. Not only does the obesogenic-environment thesis reinforce a supply-side focus with little thought as to how that might affect real estate values in ways that could replicate class and race inequalities. It also reinforces the idea that people of low socioeconomic status are somehow responsible for obesity, rather than recognizing how weight discrimination affects their class status. Finally, it neglects the possibility that chemicals are remaking bodies in serious ways, regardless of whether they contribute to obesity. Given the evidence for a much more complex etiology of obesity than too much food and too little sidewalk, the ‘doable’ interventions may be doing more harm than good.

Acknowledgements. Some of the research discussed herein was supported by the University of California at Santa Cruz’s Committee on Research. JP Jones and Brian Fulfrost supplied useful instruction on spatial analysis. Members of the University of California’s Multi-campus Research Program on Food and the Body, and three anonymous reviewers provided generous comments on earlier versions of this paper. The stunning Mesa Refuge provided a lovely space for me to work on the manuscript. I am thankful for all of this support I have received.

References Alwitt L, Donley T, 1997, “Retail stores in poor urban neighborhoods” Journal of Consumer Affairs

31 139–164 Arble D M, Bass J, Laposky A D, Vitaterna M H, Turek F W, 2009, “Circadian timing of food intake

contributes to weight gain” Obesity 17 2100–2102 Babey S H, Diamant A L, Hastert T A, Harvey S, 2008, “Designed for disease: the link between local

food environments and obesity and diabetes”, UCLA Center for Health Policy Research, http://escholarship.org/uc/item/7sf9t5wx

Bagwell S, 2011, “The role of independent fast-food outlets in obesogenic environments: a case study of East London in the UK” Environment and Planning A 43 2217–2236

Baillie-Hamilton P, 2002, “Chemical toxins: a hypothesis to explain the global obesity epidemic” Journal of Alternative and Complementary Medicine 8 185–192

Problematizing the obesogenic environment thesis 155

Björntorp P, 2001, “Do stress reactions cause abdominal obesity and comorbidities?” Obesity Reviews 2(2) 73–86

Black J L, Macinko J, 2008, “Neighborhoods and obesity: the roles of area socio-demographics, food availability and the physical environment” Nutrition Reviews 66(1) 2-20

Block J P, Scribner R A, DeSalvo K B, 2004, “Fast food, race/ethnicity, and income: a geographic analysis” American Journal of Preventive Medicine 27 211–217

Booth K M, Pinkston M M, Poston W S C, 2005, “Obesity and the built environment” Journal of the American Dietetic Association 105 S110–S117

Bray M S, Young M E, 2007, “Circadian rhythms in the development of obesity: potential role for the circadian clock within the adipocyte” Obesity Reviews 8 169–181

Brug J, van Lenthe F J, Kremers S P J, 2006, “Revisiting Kurt Lewin: how to gain insight into environmental correlates of obesogenic behaviors” American Journal of Preventive Medicine 31 525–529

California Health Interview Survey, 2006, “Counties of California: overweight and obese, BMI 25 or greater”, http://www.chis.ucla.edu/ber/county.asp?tableID=13&sort=pct

Chan S, 2009, “Data show Manhattan is svelte and Bronx is chubby, chubby” The New York Times 21 July

Cheadle A, Psaty B M, Curry S, Wagner E, Diehr P, Koepsell T, Kristal A, 1991, “Community-level comparisons between the grocery store environment and individual dietary practices” Preventive Medicine 20 250–261

Colls R, Evans B, 2009, “Introduction: questioning obesity politics” Antipode 41 1011–1020 Cotterill R W, Franklin A W, 1995,“The urban grocery store gap”, Food Marketing Policy Center,

University of Connecticut Storrs, CT Courtemanche C, Carden A, 2010, “Supersizing supercenters? The impact of wal-mart supercenters

on body mass index and obesity” SSRN, http://ssrn.com/abstract=1263316 Crossley N, 2004, “Fat is a sociological issue: obesity rates in late modern, ‘body-conscious’

societies” Social Theory and Health 2 222–253 Cummins S, Macintyre S, 2002, “ ‘Food deserts’—evidence and assumption in health policy

making” British Medical Journal 325(7361) 436–438 Cummins S, Macintyre S, 2006, “Food environments and obesity—neighbourhood or nation?”

International Journal of Epidemiology 35 100–104 Cummins S, Macintyre S, Davidson S, Ellaway A, 2005, “Measuring neighbourhood social and

material context: generation and interpretation of ecological data from routine and non-routine sources” Health and Place 11 249–260

Cummins S, Curtis S, Diez-Roux A V, Macintyre S, 2007, “Understanding and representing ‘place’ in health research: a relational approach” Social Science and Medicine 65 1825–1838

Curtis S, Riva M, 2010, “Health geographies I: complexity theory and human health” Progress in Human Geography 34 215–223

Dolinoy D C, Jirtle R L, 2008, “Environmental epigenomics in human health and disease” Environmental and Molecular Mutagenesis 49 4–8

Eisenhauer E, 2001, “In poor health: supermarket redlining and urban nutrition” GeoJournal 53 125–133

Evans B, 2006, “ ‘Gluttony or sloth’: critical geographies of bodies and morality in (anti)obesity policy” Area 38 259–267

Evans B, Crookes L, Coaffee J, 2012, “Obesity/fatness and the city: critical urban geographies” Geography Compass 6 100–110

Ewing R, Schmid T, Killingsworth R, Zlot A, Raudenbush S, 2003, “Relationship between urban sprawl and physical activity, obesity, and morbidity” American Journal of Health Promotion 18 47–57

Faulk C, Dolinoy D C, 2011, “Timing is everything: the when and how of environmentally induced changes in the epigenome of animals” Epigenetics 6 791–797

Forsyth T, 2003 Critical Political Ecology: The Politics of Environmental Science (Routledge, London)

156 J Guthman

Frank L D, Engelke P O, Schmid T L, 2003 Health and Community Design: The Impact of the Built Environment on Physical Activity (Island Press, Washington, DC)

Gard M, Wright J, 2005 The Obesity Epidemic: Science, Morality, and Ideology (Routledge, London)

Giskes K, van Lenthe F, Avendano-Pabon M, Brug J, 2011, “A systematic review of environmental factors and obesogenic dietary intakes among adults: are we getting closer to understanding obesogenic environments?” Obesity Reviews 12 e95–e106

Gladen B C, Bagan N B, Rogan W J, 2000, “Pubertal growth and development and prenatal and lactational exposure to polychlorinated biphenyls and dichlorodiphenyl dichloroethene” The Journal of Pediatrics 136 490–496

Grun F, Blumberg B, 2006, “Environmental obesogens: organotins and endocrine disruption via nuclear receptor signaling” Endocrinology 147(6) s50–s55

Grun F, Blumberg B, 2009, “Minireview: the case for obesogens” Molecular Endocrinology 23 1127–1134

Grun F, Watanabe H, Zamanian Z, Maeda L, Arima K, Cubacha R, Gardiner D M, Kanno J, Iguchi T, Blumberg B, 2006, “Endocrine-disrupting organotin compounds are potent inducers of adipogenesis in vertebrates” Molecular Endocrinology 20 2141–2155

Guthman J, 2011 Weighing in: Obesity, Food Justice, and the Limits of Capitalism (University of California Press, Berkeley, CA)

Hajer M A, 1995 The Politics of Environmental Discourse: Ecological Modernization and the Policy Process (Oxford University Press, New York)

Hatch E E, Nelson J W, Stahlhut R W, Webster T F, 2010, “Association of endocrine disruptors and obesity: perspectives from epidemiological studies” International Journal of Andrology 33 324–332

Hemphill E, Raine K, Spence J C, Smoyer-Tomic K E, 2008, “Exploring obesogenic food environments in Edmonton, Canada: the association between socioeconomic factors and fast- food outlet access” American Journal of Health Promotion 22 426–432

Herrick C, 2007, “Risky bodies: public health, social marketing and the governance of obesity” Geoforum 38 90–102

Herrick C, 2008, “To the west and east of Interstate-35: obesity, philanthropic entrepreneurialism, and the delineation of risk in Austin, Texas” Environment and Planning A 40 2715–2733

Hill J O, Peters J C, 1998, “Environmental contributions to the obesity epidemic” Science 280(5368) 1371–1374

Jasanoff S, 2004 States of Knowledge: The Co-production of Science and Social Order (Routledge, London)

Jeffery R, Baxter J, McGuire M, Linde J, 2006, “Are fast food restaurants an environmental risk factor for obesity?” International Journal of Behavioral Nutrition and Physical Activity 3 2

Joshu C E, Boehmer T K, Brownson R C, Ewing R, 2008, “Personal, neighbourhood and urban factors associated with obesity in the United States” Journal of Epidemiology and Community Health 62 202–208

Julier A, 2008, “The political economy of obesity: the fat pay all”, in Food and Culture: A Reader Eds C Counihan, P V Esterik (Routledge, New York) pp 482–499

Kirkland A, 2010, “The environmental account of obesity: a case for feminist skepticism” Signs: a Journal of Women in Culture and Society 35 463–486

Kuzawa C W, Sweet E, 2009, “Epigenetics and the embodiment of race: developmental origins of US racial disparities in cardiovascular health” American Journal of Human Biology 21 2–15

Lake A, Townshend T, 2006, “Obesogenic environments: exploring the built and food environments” Journal of the Royal Society for the Promotion of Health 126(6) 262–267

Lee H, 2006, “Obesity among California adults: racial and ethnic differences”, Public Policy Institute of California http://www.ppic.org/content/pubs/report/R_906HLR.pdf

Lopez R, 2004, “Urban sprawl and risk for being overweight or obese” American Journal of Public Health 94 1574–1579

Problematizing the obesogenic environment thesis 157

Li F, Harmer P A, Cardinal B J, Bosworth M, Acock A, Johnson-Shelton D, Moore J M, 2008, “Built environment, adiposity, and physical activity in adults aged 50–75” American Journal of Preventive Medicine 35 38–46

Li F, Harmer P, Cardinal B, Bosworth M, Johnson-Shelton D, 2009, “Obesity and the built environment: does the density of neighborhood fast-food outlets matter?” American Journal of Health Promotion 23 203–209

Lopez-Zetina J, Lee H, Friis R, 2006, “The link between obesity and the built environment: evidence from an ecological analysis of obesity and vehicle miles of travel in California” Health and Place 12 656–664

Lovasi G S, Neckerman K M, Quinn J W, Weiss C C, Rundle A, 2008, “Individual or neighborhood disadvantage modifi es the association between neighborhood walkability and body mass index” American Journal of Public Health 99 279–284

Maddock J, 2004, “The relationship between obesity and the prevalence of fast food restaurants: state-level analysis” American Journal of Health Promotion 19 137–143

Massey D S, Denton N A, 1998 American Apartheid: Segregation and the Making of the Underclass (Harvard University Press, Cambridge, MA)

Masuno H, Kidani T, Sekiya K, Sakayama K, Shiosaka T, Yamamoto H, Honda K, 2002, “Bisphenol a in combination with insulin can accelerate the conversion of 3t3-l1 fi broblasts to adipocytes” Journal of Lipid Research 43 676–684

Morland K, Wing S, Diez Roux A, Poole C, 2002, “Neighborhood characteristics associated with the location of food stores and food service places” American Journal of Preventive Medicine 22 23–29

Nelson N M, Woods C B, 2009, “Obesogenic environments: are neighbourhood environments that limit physical activity obesogenic?” Health and Place 15 917–924

Newbold R, Padilla-Banks E, Jefferson W, Heindel J, 2008, “Effects of endocrine disruptors on obesity” International Journal of Andrology 31 201–208

Papas M A, Alberg A J, Ewing R, Helzlsouer K J, Gary T L, Klassen A C, 2007, “The built environment and obesity” Epidemiologic Reviews 29 129–143

Pearce J, Blakely T, Witten K, Bartie P, 2007, “Neighborhood deprivation and access to fast-food retailing: a national study” American Journal of Preventative Medicine 32 375–382

Pearce J, Hiscock R, Blakely T, Witten K, 2009, “A national study of the association between neighbourhood access to fast-food outlets and the diet and weight of local residents” Health and Place 15 193–197

Plantinga A J, Bernell S, 2005, “A spatial economic analysis of urban land use and obesity” Journal of Regional Science 45 473–492

Poortinga W, 2006, “Perceptions of the environment, physical activity, and obesity” Social Science and Medicine 63 2835–2846

Procter K L, Clarke G P, Ransley J K, Cade J, 2008, “Micro-level analysis of childhood obesity, diet, physical activity, residential socioeconomic and social capital variables: where are the obesogenic environments in Leeds?” Area 40 323–340

Puhl R, Brownell K D, 2001, “Bias, discrimination, and obesity” Obesity 9 788–805 Raja S, Li Y, Roemmich J, Changxing M, Epstein L, Yadav P, Ticoalu A B, 2010, “Food

environment, built environment, and women’s BMI: evidence from Erie County, New York” Journal of Planning Education and Research 29 444–460

Reardon J, 2005 Race to the Finish: Identity and Governance in an Age of Genomics (Princeton University Press, Princeton, NJ)

Rosenberger R S, Sneh Y, Phipps T T, 2005, “A spatial analysis of linkages between health care expenditures, physical inactivity, obesity and recreation supply” Journal of Leisure Research 37 216–235

Saelens B E, Sallis J F, Black J B, Chen D, 2003, “Neighborhood-based differences in physical activity: an environment scale evaluation” American Journal of Public Health 93 1552–1558

Schafer Elinder L, Jansson M, 2009, “Obesogenic environments: aspects on measurement and indicators” Public Health Nutrition 12 307–315

158 J Guthman

Schein R H, 2006, “Race and landscape in the United States”, in Landscape and Race in the United States Ed. R H Schein (Routledge, New York) pp 1–21

Schrag P, 1998 Paradise Lost: California’s Experience, America’s Future (University of California Press, Berkeley, CA)

Short A, Guthman J, Raskin S, 2007, “Food deserts, oases, or mirages? Small markets and community food security in the San Francisco bay area” Journal of Planning Education and Research 26 352–364

Smith N, 1996 The New Urban Frontier: Gentrifi cation and the Revanchist City (Routledge, London) Smyth F, 2008, “Medical geography: understanding health inequalities” Progress in Human

Geography 32 119–127 Spence J, Cutumisu N, Edwards J, Raine K, Smoyer-Tomic K, 2009, “Relation between local food

environments and obesity among adults” BMC Public Health 9(1) 192, doi: 10.1186/1471-2458-9-192

Sui D Z, 2003, “Musings on the fat city: are obesity and urban forms linked?” Urban Geography 24 75–84

Swinburn B, Egger G, 2004, “The runaway weight gain train: too many accelerators, not enough brakes” British Medical Journal 329(7468) 763–739

Swinburn B, Egger G, Raza F, 1999, “Dissecting obesogenic environments: the development and application of a framework for identifying and prioritizing environmental interventions for obesity” Preventative Medicine 29 563–570

Thayer Z, Kuzawa C, 2011, “Biological memories of past environments: epigenetic pathways to health disparities” Epigenetics 6 798–803

Townshend T, Lake A A, 2009, “Obesogenic urban form: theory, policy and practice” Health and Place 15 909–916

Tunstall H V Z, Shaw M, Dorling D, 2004, “Places and health” Journal of Epidemiology and Community Health 58 6–10

USDA, 2002 Consumption in America, Agriculture Factbook 2001-2002 US Department of Agriculture, Washington DC, http://www.usda.gov/factbook

USDA, 2010, “What we eat in America”, US Department of Agriculture, http://www.ars.usda.gov/Services/docs.htm?docid=18349#

Vandegrift D, Yoked T, 2004, “Obesity rates, income, and suburban sprawl: an analysis of US states” Health and Place 10 221–229

Wang M C, Cubbin C, Ahn D, Winkleby M A, 2008, “Changes in neighbourhood food store environment, food behaviour and body mass index, 1981–1990” Public Health Nutrition 11 963–970

© 2013 Pion and its Licensors