摘要：Objectives. We estimated the association between obesity and features of the natural environment. We asked whether the association is mediated by diet or by physical activity. Methods. Using county-level data from the contiguous United States, we regressed adult obesity prevalence on 9 measures of the natural environment. Our regression model accounted for spatial correlation, and controlled for county demographics and the built environment. We included physical activity and diet (proxied by food purchases) as potential mediators. Results. Obesity was more prevalent in counties that are hot in July or cold in January. To a lesser degree, obesity was more prevalent in counties that are dark in January or rainy (but not snowy) year-round. Other aspects of the natural environment—including wind, trees, waterfront, and hills and mountains—had little or no association with obesity. Nearly all of the association between obesity and the natural environment was mediated by physical activity; none was mediated by diet. Conclusions. Hot summers and cold winters appear to promote obesity by discouraging physical activity. Attempts to encourage physical activity should compensate for the effects of extreme temperatures. Efforts to understand and combat the obesity epidemic often focus on the “built environment.” Features of the built environment—such as parks, recreational facilities, 1 and dense mixed-use neighborhoods where residents can walk to stores and offices 2 —encourage physical activity, 2,3 although their effect on obesity is not as clear. 4 By contrast, sprawling built environments, with little public space and most destinations accessible only by car, discourage physical activity 5 and promote sedentary indoor habits that can lead to obesity. 5,6 Increases in sprawl have been blamed for one tenth of the rise in US obesity prevalence, 7 although the process of sprawl began decades before 8 obesity prevalence began to rise in the 1980s. 9 In this article we focus not on the built environment, but on obesity’s relationship with the natural environment. Testing this association is of interest for several reasons. First, it is a test of the broader hypothesis that obesity is affected by the physical environment—a hypothesis that is fundamental to work on the built environment. Second, features of the built environment can only encourage physical activity within the constraints and opportunities afforded by the natural environment. For example, a sidewalk may get little use where the summer is very hot or the winter is very cold, and a bicycle path may get less use where the land is steep than where it is flat. Another reason to study the natural environment is that for decades the US population has been moving toward western and southwestern regions where the climate is warmer and drier, with less difference between summer and winter, than the climate in the north and east. 10 In addition, the US climate has grown warmer and less seasonal as a part of global climate change. 10 The combined effect of these changes on the temperature experienced by most people in the United States began to accelerate after 1980, 10 around the time that obesity prevalence began to rise. 9 It would be helpful to know whether changes in the US natural environment have speeded or slowed the progress of the obesity epidemic. There are several mechanisms through which the natural environment might affect obesity. The most obvious mechanism is physical activity, which varies in response to the weather. 11 Adults and children are most active between 15°C and 20°C (60°F and 70°F), 5 so that residents of cold climates are most active in spring and summer, residents of warm climates are most active in fall and winter, 11 and residents of mild climates can be active year-round. Activity also increases in response to sunlight and decreases in response to wind or rain. 5,12,13 Snow may have different effects on different individuals, encouraging outdoor activity by the fit but not by the obese. 14 Some studies have reported that individuals are more physically active in areas with hills or enjoyable scenery. 15,16 In addition to affecting physical activity, one aspect of the natural environment—temperature—is also related to energy balance and body shape. Exposure to cold temperatures induces behavioral and physiological changes that generate or conserve heat. Short-term exposure to cold temperatures increases food intake and metabolic energy expenditure. 17–19 Long-term exposure to cold can lead to the accumulation of insulating fat and a larger, rounder body shape that reduces heat loss by increasing the ratio of body volume to surface area. The tendency for cold-weather mammals to have a larger volume-to-area ratio is known in ecology as Bergmann’s rule. 20 Bergmann’s rule was used to interpret studies of US soldiers in the 1940s and 1950s, which found that soldiers who were raised in colder US states were heavier, with a larger body mass index (BMI; defined as weight in kilograms divided by the square of height in meters [kg/m2]) and a larger volume-to-area ratio, than soldiers raised in warmer states. 21 Likewise, soldiers who were assigned to serve in cold climates consumed more calories and grew heavier than soldiers who served in the tropics. 22 Today evidence for Bergmann’s rule is much harder to see in the US population. Contrary to Bergmann’s rule, residents of the warm southeast are now heavier than residents of the cool northeast and mountain west. One possible reason for the changing relationship between temperature and body size is the spread of residential air conditioning. Since 1960 the installation of residential air conditioning in the United States has increased from 13% to 87% of homes, 23,24 so that more individuals can now escape heat and humidity by retreating indoors. Indoor heating has also improved; in cold climates, indoor winter temperatures have risen for decades and are now close to the “thermoneutral zone” (25°C to 27°C or 77°F to 81°F) where metabolic rates are lowest. 25 What this may imply is that residents of extreme climates now spend more of their time indoors 26 in temperatures where metabolic rates are rarely elevated and there is ample opportunity for sedentary behavior and snacking. 27 In the 1950s it was speculated that air conditioning would change the relationship between temperature and body size, and more recently it has been suggested that indoor climate control has contributed something to the obesity epidemic. 26,28 It should be acknowledged, however, that the spread of residential air conditioning in the United States started decades 23 before obesity prevalence began to rise in the 1980s. 9 In this study, we estimated the association between obesity prevalence and various aspects of the natural environment. Temperature—both winter and summer—is the natural-environment variable whose relationship to obesity is most strongly motivated by past research, but we also estimated the association between obesity prevalence and winter sunlight, summer humidity, rain, snow, wind, forests, waterfront, and topography (i.e., plains, hills, mountains). The relationship between obesity and the natural environment has been studied before, 29–32 but our study makes some advances. First, ours is the first modern study to estimate the association between obesity and individual measures of the natural environment. Past studies relied instead on a summary scale of “natural amenity” 29,30,32 or “climate amenity,” 31 which collapsed into 1 scale a variety of natural environment measures selected from among temperature, humidity, wind, rain, snow, sunlight, topography, water area, forest cover, and tourism. From these single-scale studies it is hard to tell which features of the natural environment are most associated with obesity. Our study clarifies which aspects of the natural environment matter most. Our study also improves on its predecessors by considering a wider range of mediating variables. Past studies have typically assumed that the natural environment affects obesity by encouraging or discouraging physical activity. 30–32 Yet one study estimated that physical activity mediated only 15% of the association between obesity and the natural environment, 32 and our previous discussion suggests that diet and metabolism could also play a mediating role. In our study, we considered both diet and physical activity as potential mediators. We would have liked to have considered metabolism as well, but metabolic data are unavailable at the population level.
