A recent study purporting to debunk the link between sprawl and ill health has gotten much publicity recently. Some commentators have trotted out the old argument that plenty of city-dwellers, especially in poor areas, are fat, so therefore sprawl really just doesn't matter.
I am perfectly willing to concede for the sake of argument that poorer people tend to be fatter (though a CDC report suggests that this correlation is more modest than some believe). But this doesn't mean that sprawl doesn't matter, any more than the strong link between smoking and lung cancer disproves the existence of other causes of lung cancer such as air pollution.
If sprawl was truly irrelevant to obesity, rich urbanites would be no thinner than rich suburbanites, and poor urbanites would be no thinner than poor suburbanites. But county-level data (available here) suggests otherwise.
I began by looking at New York, because in that region even the central city itself is divided into different counties; thus, we can distinguish between areas of varying density in a way we cannot in regions where both the city and suburbs are in the same county.
In the least car-oriented county, New York County (also known as Manhattan) the obesity rate is only 15.1%. By contrast, suburban counties have lower poverty rates, yet have higher obesity rates. Nassau County has only a 5.2% poverty rate (less than one-third that of Manhattan), but a 21.3% obesity rate- a 16 point gap. Exurban Suffolk County had an even bigger gap, with a poverty rate comparable to that of Nassau (5.7%) but a higher obesity rate (25.6%). Exurban Putnam County had a 5.4% poverty rate and a obesity rate of 28.7%.
What if you compare these suburbs to New York's poorer urban boroughs? Even those counties' obesity rates are comparable to those of the rich suburbs. Both Brooklyn and the Bronx have poverty rates over 20%- yet Brooklyn's obesity rate is 24.5%, and the Bronx's obesity rate is 28.4%- only slightly higher than that of Suffolk County, and lower than that of Putnam County.
In other words, not only are New York's low-poverty suburbs fatter than New York's most affluent urban county (Manhattan), but they are fatter than Brooklyn, and almost as fat as the poorer-still Bronx. So in places where most people don't own cars, the positive effects of urban life may cancel out the harmful impacts of poverty, at least regarding obesity.
But how do we measure the obesity/poverty relationship in places where urban counties are not so walkable and transit-oriented as Manhattan? The measure I chose was to compare the gap between a county's obesity rate and its poverty rate. If, as in New York City's counties, a place's obesity rate is as low as its poverty rate, its people are relatively thin controlling for poverty. But if, as in New York's suburbs, the obesity rate is far higher than the poverty rate, the people are fatter than their paychecks.
In particular, I focused on three regions where the city is its own county, so that I could distinguish city from suburb. I started with St. Louis, which has an exceptionally poor central city. In St. Louis, the poverty rate is 26% while the obesity rate is 33.9% - a 7.9% gap. In next-door St. Louis County, the poverty rate is much lower (9.7%) while the obesity rate is almost as high as in the city (29%)- a 19.3% gap. And in well-off exurban St. Charles County, the poverty rate is even lower (4.9%) while the obesity rate is comparable to St. Louis County (29.7%)- almost a 25 point gap. In other words, as you move out from the urban core the poverty rate nosedives while the obesity rate declines only slightly- evidence, it seems to me, that if St. Louis city was richer it would probably be thinner than its suburbs, or if St. Louis County had as many poor people as St. Louis its' residents auto-dependent lifestyle would make them fatter still.
Then I looked at Philadelphia, which had a somewhat similar pattern; the city's obesity rate is higher than that of its suburbs (31.5% for the city, 21-27% for Montgomery, Chester, Delaware, and Bucks Counties). But the poverty/obesity gap is much smaller for the city. In Philadelphia, the poverty rate is 25.6% and the obesity rate is 31.5%- a 5.9% gap. By contrast, in the suburban counties the poverty rate ranged from 5.1% (Chester) to 9.5% (Delaware), yielding gaps of about 14-20%. For example, exurban Chester County had a obesity rate of 21.2% and a poverty rate of only 6.7%, a 14.5% gap.
What about where the central city is a bit more affluent? San Francisco is also its own county, and has only a slightly higher poverty rate than most suburban counties. (The city's poverty rate is 12.3 percent; Marin and San Mateo Counties have poverty rates around 7%, and Contra Costa County's poverty rate is around 10%). Of these three suburban counties, two (San Mateo and Contra Costa) have obesity rates higher than San Francisco, despite having less poverty. Marin has less obesity than San Francisco, but even here the poverty/obesity gap is lower in the city. San Francisco has 12.3% poverty and 17.1% obesity (a 4.8% gap) and Marin has 7.2% poverty and 15.3% obesity (an 8.1% gap).
In sum, it seems to me that when one controls for poverty, obesity is concentrated in suburbia. One might argue that this correlation is not related to causation: that people who live in suburbs are the sort of people who prefer poor diets and lack of exercise. I do not know how one would prove or disprove this proposition; however, I did find a map listing fruit and vegetable consumption by county. If urbanites generally preferred healthier lifestyles, they might consume more fruits and vegetables than suburbanites.
But in some regions, city residents consumed even fewer fruits and vegetables than suburbanites; for example, in St. Louis, 79 percent of people did not eat enough, as opposed to 74 percent of St. Louis County residents. Even in places where city-dwellers were thinner than suburbanites, fruit and vegetable consumption was similar in city and suburb. For example, 74 percent of Manhattanites did not eat enough, about the same as in Nassau County (73) and Suffolk (74).
Thus, it does not appear that suburbanites are otherwise more or less health-conscious than city-dwellers- a fact that reinforces the link between suburban automobile-dependency and obesity.