Archive for the ‘Methodology’ Category

Do obese people choose obese environments?

Thursday, April 29th, 2010

A continuing issue regarding the influence of the built environment on health has been  whether certain types of environments: sprawled metropolitan areas or unwalkable neighborhoods cause obesity or whether obese people simply choose these areas because they support a low physical activity lifestyle or make it easier for obese people to live there.  Which came first: obesity or the neighborhood?

Part of the reason for this uncertainty is the nature of the evidence.  Most of it comes from cross-sectional data (individuals are asked questions at a single point in time). A major limitation of these types of studies is that no conclusions about the directionality of the associations can be made.

There have been some attempts to use cohort data, information on a set of individuals collected at several or more time intervals.  The problem with these datasets is that they aren’t very many of them and they tend to be small.  Small studies may lack the statistical power to identify the subtle, but important, effects of sprawl on obesity, for example.

One of the few longitudinal datasets that can be used to study the built environment is the National Longitudinal Survey of Youth (NLSY) which consists of two national samples:  persons who were born between 1957 and 1965 and the natural born children of mothers in the original cohort  (a third cohort has been established but its oldest members are only 26, perhaps not yet old enough fo use in an obesity study).  These cohorts were selected in order to enable study of how people entered the labor force and progressed their careers. These cohort has the essential features that are needed for a health and the environment study: place of residence (not publicly available but accessible upon special request), data on height and weight (collected to assess disability status) and data collected almost every other year since the cohort’s start date.

Reid Ewing used this dataset, in conjunction with his urban sprawl measure (developed in conjunction with Smart Growth America) to test the which came first problem.  What he found was that the built environment (sprawl) did not seem to be associated with obesity in longitudinal analyses.  Note that this does not directly put to rest which came first, only that the NLSY data doesn’t support the sprawl to obesity hypothesis.  But the NLSY may be too small (only about 10,000 persons) to uncover the association.  So the controversy continues.

There has been a great deal of research on residential choice.  Economists and urban planners have been curious about this for years.  They tend to find that affordability and school quality are the most important predictive factors for how a household chooses its location.  Access to jobs is also important.  These studies never included any data on walkability or urban design.  They never considered that people would choose neighborhoods based on obesity status.

So as of this time, we cannot determine which came first:  obesity or the neighborhood.

Health vs. design professions: Different disciplines, different methodologies

Thursday, April 15th, 2010

Connecting the design professions (urban planning, urban design, architecture, transportation planning and civil engineering) with the health professions (public health, medicine, epidemiology) is not easy.  They speak different languages and, even more important, use different methods for deciding what best describes reality and how it might best be improved.  It is not that one is superior to the other. They are simply different.

When my colleague, Greg Howard, and I were putting together our course on the built environment and health, we had to submit the course syllabi to the school’s curriculum committee, an important exercise through which course goals are sharpened and the subject matter is made more encompassing for the topic on hand.  When we developed our syllabus, we had no other courses to use as a model.  Fortunately Nisha Botchway has published a model built environment and health curriculum that others can use to develop their own course. (

The curriculum committee wanted us to use published journal articles that used standard epidemiological methods to assess features of the built environment.  Where were the case control studies on the proper setback of buildings from the street?  Why didn’t we include cohort studies that Modernists used to design their ideal communities?  They were surprised to learn how few features of the built environment have been subjected to epidemiological testing.

In my courses and lectures, I have been fortunate to have planners and designers who have expressed frustration  with how  strangely health research proceeds.  If there is a potential problem, why don’t we just send in a committee of experts to examine  the situation and have them make recommendations for modifications in building practices.  Really, why try to find a sample of thousands of people living in different living environments?  Isn’t that inefficient and unnecessary?

In health, the methods are case control and cohort studies.  If those can’t be done, maybe a case study or ecologic study might suffice, but there is an unease with the evidence.  Health researchers would really want a double blinded study of features of the built environment but that is technically, financially and morally impossible.  No single study is sufficient, typically health people want to see multiple studies looking at the problem using differing approaches before a definitive decision is  made.

In building design, typically changes are made after there is some sort of catastrophic failure.  Fire codes changed after the Station Nightclub fire in Rhode Island.  The recent earthquake in Chile resulted in engineers going out to towns near the epicenter to determine why buildings collapsed.  As a result, codes will be modified from Seattle to Istanbul. These methods are relatively quick and ultimately work to protect human life.

Some researchers have crossed over from there beginning disciplines to use the methods of the other.  Robert Cervero’s studies of travel behavior and Reid Ewing’s sprawl studies are excellent examples of the possibilities.   However, for the most part, the two disciplines use different methods.  Both could learn from the other.

Again, neither set of methods can be said to be superior to the other.  Perhaps the answer is that each discipline should try to adapt the methods of the other.  Maybe send in engineers to understand the potential impacts of PCBDs on health or use epidemiological studies of single family homes to determine which designs best promote healthy living.  We can learn from each other.